Support for media extraction from .mkv/.mka Matroska files in stagefright.

Change-Id: I4c26579828ad575523ccf58b0b5cb144046c04ca
related-to-bug: 2483739
diff --git a/include/media/stagefright/MediaDefs.h b/include/media/stagefright/MediaDefs.h
index 207195a..a9a0d55 100644
--- a/include/media/stagefright/MediaDefs.h
+++ b/include/media/stagefright/MediaDefs.h
@@ -22,6 +22,7 @@
 
 extern const char *MEDIA_MIMETYPE_IMAGE_JPEG;
 
+extern const char *MEDIA_MIMETYPE_VIDEO_VPX;
 extern const char *MEDIA_MIMETYPE_VIDEO_AVC;
 extern const char *MEDIA_MIMETYPE_VIDEO_MPEG4;
 extern const char *MEDIA_MIMETYPE_VIDEO_H263;
@@ -38,6 +39,7 @@
 extern const char *MEDIA_MIMETYPE_CONTAINER_MPEG4;
 extern const char *MEDIA_MIMETYPE_CONTAINER_WAV;
 extern const char *MEDIA_MIMETYPE_CONTAINER_OGG;
+extern const char *MEDIA_MIMETYPE_CONTAINER_MATROSKA;
 
 }  // namespace android
 
diff --git a/media/java/android/media/MediaFile.java b/media/java/android/media/MediaFile.java
index 9d1d420..f1da72f 100644
--- a/media/java/android/media/MediaFile.java
+++ b/media/java/android/media/MediaFile.java
@@ -46,8 +46,9 @@
     public static final int FILE_TYPE_WMA     = 6;
     public static final int FILE_TYPE_OGG     = 7;
     public static final int FILE_TYPE_AAC     = 8;
+    public static final int FILE_TYPE_MKA     = 9;
     private static final int FIRST_AUDIO_FILE_TYPE = FILE_TYPE_MP3;
-    private static final int LAST_AUDIO_FILE_TYPE = FILE_TYPE_AAC;
+    private static final int LAST_AUDIO_FILE_TYPE = FILE_TYPE_MKA;
 
     // MIDI file types
     public static final int FILE_TYPE_MID     = 11;
@@ -63,8 +64,9 @@
     public static final int FILE_TYPE_3GPP2   = 24;
     public static final int FILE_TYPE_WMV     = 25;
     public static final int FILE_TYPE_ASF     = 26;
+    public static final int FILE_TYPE_MKV     = 27;
     private static final int FIRST_VIDEO_FILE_TYPE = FILE_TYPE_MP4;
-    private static final int LAST_VIDEO_FILE_TYPE = FILE_TYPE_ASF;
+    private static final int LAST_VIDEO_FILE_TYPE = FILE_TYPE_MKV;
     
     // Image file types
     public static final int FILE_TYPE_JPEG    = 31;
@@ -134,6 +136,7 @@
         addFileType("OGG", FILE_TYPE_OGG, "application/ogg");
         addFileType("OGA", FILE_TYPE_OGG, "application/ogg");
         addFileType("AAC", FILE_TYPE_AAC, "audio/aac");
+        addFileType("MKA", FILE_TYPE_MKA, "audio/x-matroska");
  
         addFileType("MID", FILE_TYPE_MID, "audio/midi");
         addFileType("MIDI", FILE_TYPE_MID, "audio/midi");
@@ -151,6 +154,8 @@
         addFileType("3GPP", FILE_TYPE_3GPP, "video/3gpp");
         addFileType("3G2", FILE_TYPE_3GPP2, "video/3gpp2");
         addFileType("3GPP2", FILE_TYPE_3GPP2, "video/3gpp2");
+        addFileType("MKV", FILE_TYPE_MKV, "video/x-matroska");
+        addFileType("WEBM", FILE_TYPE_MKV, "video/x-matroska");
         if (isWMVEnabled()) {
             addFileType("WMV", FILE_TYPE_WMV, "video/x-ms-wmv");
             addFileType("ASF", FILE_TYPE_ASF, "video/x-ms-asf");
diff --git a/media/libstagefright/Android.mk b/media/libstagefright/Android.mk
index e54fb67..681943f 100644
--- a/media/libstagefright/Android.mk
+++ b/media/libstagefright/Android.mk
@@ -75,7 +75,8 @@
         libstagefright_avcdec \
         libstagefright_m4vh263dec \
         libstagefright_mp3dec \
-        libstagefright_vorbisdec
+        libstagefright_vorbisdec \
+        libstagefright_matroska \
 
 LOCAL_SHARED_LIBRARIES += \
         libstagefright_amrnb_common \
diff --git a/media/libstagefright/DataSource.cpp b/media/libstagefright/DataSource.cpp
index a66f86b..475422d 100644
--- a/media/libstagefright/DataSource.cpp
+++ b/media/libstagefright/DataSource.cpp
@@ -20,6 +20,8 @@
 #include "include/WAVExtractor.h"
 #include "include/OggExtractor.h"
 
+#include "matroska/MatroskaExtractor.h"
+
 #include <media/stagefright/CachingDataSource.h>
 #include <media/stagefright/DataSource.h>
 #include <media/stagefright/FileSource.h>
@@ -94,6 +96,7 @@
     RegisterSniffer(SniffAMR);
     RegisterSniffer(SniffWAV);
     RegisterSniffer(SniffOgg);
+    RegisterSniffer(SniffMatroska);
 }
 
 // static
diff --git a/media/libstagefright/MediaDefs.cpp b/media/libstagefright/MediaDefs.cpp
index 4b3813b..327a0ce 100644
--- a/media/libstagefright/MediaDefs.cpp
+++ b/media/libstagefright/MediaDefs.cpp
@@ -20,6 +20,7 @@
 
 const char *MEDIA_MIMETYPE_IMAGE_JPEG = "image/jpeg";
 
+const char *MEDIA_MIMETYPE_VIDEO_VPX = "video/x-vnd.on2.vp8";
 const char *MEDIA_MIMETYPE_VIDEO_AVC = "video/avc";
 const char *MEDIA_MIMETYPE_VIDEO_MPEG4 = "video/mp4v-es";
 const char *MEDIA_MIMETYPE_VIDEO_H263 = "video/3gpp";
@@ -36,5 +37,6 @@
 const char *MEDIA_MIMETYPE_CONTAINER_MPEG4 = "video/mpeg4";
 const char *MEDIA_MIMETYPE_CONTAINER_WAV = "audio/wav";
 const char *MEDIA_MIMETYPE_CONTAINER_OGG = "application/ogg";
+const char *MEDIA_MIMETYPE_CONTAINER_MATROSKA = "video/x-matroska";
 
 }  // namespace android
diff --git a/media/libstagefright/MediaExtractor.cpp b/media/libstagefright/MediaExtractor.cpp
index 513f49c..376d715 100644
--- a/media/libstagefright/MediaExtractor.cpp
+++ b/media/libstagefright/MediaExtractor.cpp
@@ -24,6 +24,8 @@
 #include "include/WAVExtractor.h"
 #include "include/OggExtractor.h"
 
+#include "matroska/MatroskaExtractor.h"
+
 #include <media/stagefright/DataSource.h>
 #include <media/stagefright/MediaDefs.h>
 #include <media/stagefright/MediaExtractor.h>
@@ -69,6 +71,8 @@
         return new WAVExtractor(source);
     } else if (!strcasecmp(mime, MEDIA_MIMETYPE_CONTAINER_OGG)) {
         return new OggExtractor(source);
+    } else if (!strcasecmp(mime, MEDIA_MIMETYPE_CONTAINER_MATROSKA)) {
+        return new MatroskaExtractor(source);
     }
 
     return NULL;
diff --git a/media/libstagefright/StagefrightMediaScanner.cpp b/media/libstagefright/StagefrightMediaScanner.cpp
index 03287dd1..ab17b04 100644
--- a/media/libstagefright/StagefrightMediaScanner.cpp
+++ b/media/libstagefright/StagefrightMediaScanner.cpp
@@ -42,7 +42,8 @@
     static const char *kValidExtensions[] = {
         ".mp3", ".mp4", ".m4a", ".3gp", ".3gpp", ".3g2", ".3gpp2",
         ".mpeg", ".ogg", ".mid", ".smf", ".imy", ".wma", ".aac",
-        ".wav", ".amr", ".midi", ".xmf", ".rtttl", ".rtx", ".ota"
+        ".wav", ".amr", ".midi", ".xmf", ".rtttl", ".rtx", ".ota",
+        ".mkv", ".mka", ".webm"
     };
     static const size_t kNumValidExtensions =
         sizeof(kValidExtensions) / sizeof(kValidExtensions[0]);
diff --git a/media/libstagefright/StagefrightMetadataRetriever.cpp b/media/libstagefright/StagefrightMetadataRetriever.cpp
index 258be74..9d89c20 100644
--- a/media/libstagefright/StagefrightMetadataRetriever.cpp
+++ b/media/libstagefright/StagefrightMetadataRetriever.cpp
@@ -378,6 +378,24 @@
     // The duration value is a string representing the duration in ms.
     sprintf(tmp, "%lld", (maxDurationUs + 500) / 1000);
     mMetaData.add(METADATA_KEY_DURATION, String8(tmp));
+
+    if (numTracks == 1) {
+        const char *fileMIME;
+        CHECK(meta->findCString(kKeyMIMEType, &fileMIME));
+
+        if (!strcasecmp(fileMIME, "video/x-matroska")) {
+            sp<MetaData> trackMeta = mExtractor->getTrackMetaData(0);
+            const char *trackMIME;
+            CHECK(trackMeta->findCString(kKeyMIMEType, &trackMIME));
+
+            if (!strncasecmp("audio/", trackMIME, 6)) {
+                // The matroska file only contains a single audio track,
+                // rewrite its mime type.
+                mMetaData.add(
+                        METADATA_KEY_MIMETYPE, String8("audio/x-matroska"));
+            }
+        }
+    }
 }
 
 
diff --git a/media/libstagefright/matroska/Android.mk b/media/libstagefright/matroska/Android.mk
new file mode 100644
index 0000000..0e72198
--- /dev/null
+++ b/media/libstagefright/matroska/Android.mk
@@ -0,0 +1,16 @@
+LOCAL_PATH:= $(call my-dir)
+include $(CLEAR_VARS)
+
+LOCAL_SRC_FILES:=                 \
+        MatroskaExtractor.cpp     \
+        mkvparser.cpp             \
+
+LOCAL_C_INCLUDES:= \
+	$(JNI_H_INCLUDE) \
+        $(TOP)/external/opencore/extern_libs_v2/khronos/openmax/include \
+
+LOCAL_CFLAGS += -Wno-multichar
+
+LOCAL_MODULE:= libstagefright_matroska
+
+include $(BUILD_STATIC_LIBRARY)
diff --git a/media/libstagefright/matroska/MatroskaExtractor.cpp b/media/libstagefright/matroska/MatroskaExtractor.cpp
new file mode 100644
index 0000000..197ccf8
--- /dev/null
+++ b/media/libstagefright/matroska/MatroskaExtractor.cpp
@@ -0,0 +1,504 @@
+/*
+ * Copyright (C) 2010 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+//#define LOG_NDEBUG 0
+#define LOG_TAG "MatroskaExtractor"
+#include <utils/Log.h>
+
+#include "MatroskaExtractor.h"
+
+#include "mkvparser.hpp"
+
+#include <media/stagefright/DataSource.h>
+#include <media/stagefright/MediaBuffer.h>
+#include <media/stagefright/MediaDebug.h>
+#include <media/stagefright/MediaDefs.h>
+#include <media/stagefright/MediaErrors.h>
+#include <media/stagefright/MediaSource.h>
+#include <media/stagefright/MetaData.h>
+#include <utils/String8.h>
+
+namespace android {
+
+struct DataSourceReader : public mkvparser::IMkvReader {
+    DataSourceReader(const sp<DataSource> &source)
+        : mSource(source) {
+    }
+
+    virtual int Read(long long position, long length, unsigned char* buffer) {
+        CHECK(position >= 0);
+        CHECK(length >= 0);
+
+        if (length == 0) {
+            return 0;
+        }
+
+        ssize_t n = mSource->readAt(position, buffer, length);
+
+        if (n <= 0) {
+            return -1;
+        }
+
+        return 0;
+    }
+
+    virtual int Length(long long* total, long long* available) {
+        off_t size;
+        if (mSource->getSize(&size) != OK) {
+            return -1;
+        }
+
+        if (total) {
+            *total = size;
+        }
+
+        if (available) {
+            *available = size;
+        }
+
+        return 0;
+    }
+
+private:
+    sp<DataSource> mSource;
+
+    DataSourceReader(const DataSourceReader &);
+    DataSourceReader &operator=(const DataSourceReader &);
+};
+
+////////////////////////////////////////////////////////////////////////////////
+
+#include <ctype.h>
+static void hexdump(const void *_data, size_t size) {
+    const uint8_t *data = (const uint8_t *)_data;
+    size_t offset = 0;
+    while (offset < size) {
+        printf("0x%04x  ", offset);
+
+        size_t n = size - offset;
+        if (n > 16) {
+            n = 16;
+        }
+
+        for (size_t i = 0; i < 16; ++i) {
+            if (i == 8) {
+                printf(" ");
+            }
+
+            if (offset + i < size) {
+                printf("%02x ", data[offset + i]);
+            } else {
+                printf("   ");
+            }
+        }
+
+        printf(" ");
+
+        for (size_t i = 0; i < n; ++i) {
+            if (isprint(data[offset + i])) {
+                printf("%c", data[offset + i]);
+            } else {
+                printf(".");
+            }
+        }
+
+        printf("\n");
+
+        offset += 16;
+    }
+}
+
+struct MatroskaSource : public MediaSource {
+    MatroskaSource(
+            const sp<MatroskaExtractor> &extractor, size_t index);
+
+    virtual status_t start(MetaData *params);
+    virtual status_t stop();
+
+    virtual sp<MetaData> getFormat();
+
+    virtual status_t read(
+            MediaBuffer **buffer, const ReadOptions *options);
+
+private:
+    enum Type {
+        AVC,
+        AAC,
+        OTHER
+    };
+
+    sp<MatroskaExtractor> mExtractor;
+    size_t mTrackIndex;
+    unsigned long mTrackNum;
+    Type mType;
+    mkvparser::Cluster *mCluster;
+    const mkvparser::BlockEntry *mBlockEntry;
+
+    status_t advance();
+
+    MatroskaSource(const MatroskaSource &);
+    MatroskaSource &operator=(const MatroskaSource &);
+};
+
+MatroskaSource::MatroskaSource(
+        const sp<MatroskaExtractor> &extractor, size_t index)
+    : mExtractor(extractor),
+      mTrackIndex(index),
+      mType(OTHER),
+      mCluster(NULL),
+      mBlockEntry(NULL) {
+    mTrackNum = mExtractor->mTracks.itemAt(index).mTrackNum;
+
+    const char *mime;
+    CHECK(mExtractor->mTracks.itemAt(index).mMeta->
+            findCString(kKeyMIMEType, &mime));
+
+    if (!strcasecmp(mime, MEDIA_MIMETYPE_VIDEO_AVC)) {
+        mType = AVC;
+    } else if (!strcasecmp(mime, MEDIA_MIMETYPE_AUDIO_AAC)) {
+        mType = AAC;
+    }
+}
+
+status_t MatroskaSource::start(MetaData *params) {
+    mCluster = NULL;
+    mBlockEntry = NULL;
+
+    return OK;
+}
+
+status_t MatroskaSource::stop() {
+    return OK;
+}
+
+sp<MetaData> MatroskaSource::getFormat() {
+    return mExtractor->mTracks.itemAt(mTrackIndex).mMeta;
+}
+
+status_t MatroskaSource::advance() {
+    for (;;) {
+        if (mBlockEntry == NULL || mBlockEntry->EOS()) {
+            if (mCluster == NULL) {
+                mCluster = mExtractor->mSegment->GetFirst();
+            } else {
+                mCluster = mExtractor->mSegment->GetNext(mCluster);
+            }
+            if (mCluster == NULL || mCluster->EOS()) {
+                return ERROR_END_OF_STREAM;
+            }
+            mBlockEntry = mCluster->GetFirst();
+        }
+
+        if (mBlockEntry->GetBlock()->GetTrackNumber() != mTrackNum) {
+            mBlockEntry = mCluster->GetNext(mBlockEntry);
+            continue;
+        }
+
+        break;
+    }
+
+    return OK;
+}
+
+status_t MatroskaSource::read(
+        MediaBuffer **out, const ReadOptions *options) {
+    *out = NULL;
+
+    int64_t seekTimeUs;
+    if (options && options->getSeekTo(&seekTimeUs)) {
+        mBlockEntry = NULL;
+        mCluster = mExtractor->mSegment->GetCluster(seekTimeUs * 1000ll);
+
+        status_t err;
+        while ((err = advance()) == OK && !mBlockEntry->GetBlock()->IsKey()) {
+            mBlockEntry = mCluster->GetNext(mBlockEntry);
+        }
+
+        if (err != OK) {
+            return ERROR_END_OF_STREAM;
+        }
+    }
+
+    if (advance() != OK) {
+        return ERROR_END_OF_STREAM;
+    }
+
+    const mkvparser::Block *block = mBlockEntry->GetBlock();
+    size_t size = block->GetSize();
+    long long timeNs = block->GetTime(mCluster);
+
+    MediaBuffer *buffer = new MediaBuffer(size + 2);
+    buffer->meta_data()->setInt64(kKeyTime, (timeNs + 500) / 1000);
+
+    long res = block->Read(
+            mExtractor->mReader, (unsigned char *)buffer->data() + 2);
+
+    if (res != 0) {
+        return ERROR_END_OF_STREAM;
+    }
+
+    buffer->set_range(2, size);
+
+    if (mType == AVC) {
+        CHECK(size >= 2);
+
+        uint8_t *data = (uint8_t *)buffer->data();
+
+        unsigned NALsize = data[2] << 8 | data[3];
+        CHECK_EQ(size, NALsize + 2);
+
+        memcpy(data, "\x00\x00\x00\x01", 4);
+        buffer->set_range(0, size + 2);
+    } else if (mType == AAC) {
+        // There's strange junk at the beginning...
+
+        const uint8_t *data = (const uint8_t *)buffer->data() + 2;
+        size_t offset = 0;
+        while (offset < size && data[offset] != 0x21) {
+            ++offset;
+        }
+        buffer->set_range(2 + offset, size - offset);
+    }
+
+    *out = buffer;
+
+#if 0
+    hexdump((const uint8_t *)buffer->data() + buffer->range_offset(),
+            buffer->range_length());
+#endif
+
+    mBlockEntry = mCluster->GetNext(mBlockEntry);
+
+    return OK;
+}
+
+////////////////////////////////////////////////////////////////////////////////
+
+MatroskaExtractor::MatroskaExtractor(const sp<DataSource> &source)
+    : mDataSource(source),
+      mReader(new DataSourceReader(mDataSource)),
+      mSegment(NULL) {
+    mkvparser::EBMLHeader ebmlHeader;
+    long long pos;
+    if (ebmlHeader.Parse(mReader, pos) < 0) {
+        return;
+    }
+
+    long long ret =
+        mkvparser::Segment::CreateInstance(mReader, pos, mSegment);
+
+    if (ret) {
+        CHECK(mSegment == NULL);
+        return;
+    }
+
+    ret = mSegment->Load();
+
+    if (ret < 0) {
+        delete mSegment;
+        mSegment = NULL;
+        return;
+    }
+
+    addTracks();
+}
+
+MatroskaExtractor::~MatroskaExtractor() {
+    delete mSegment;
+    mSegment = NULL;
+
+    delete mReader;
+    mReader = NULL;
+}
+
+size_t MatroskaExtractor::countTracks() {
+    return mTracks.size();
+}
+
+sp<MediaSource> MatroskaExtractor::getTrack(size_t index) {
+    if (index >= mTracks.size()) {
+        return NULL;
+    }
+
+    return new MatroskaSource(this, index);
+}
+
+sp<MetaData> MatroskaExtractor::getTrackMetaData(
+        size_t index, uint32_t flags) {
+    if (index >= mTracks.size()) {
+        return NULL;
+    }
+
+    return mTracks.itemAt(index).mMeta;
+}
+
+static void addESDSFromAudioSpecificInfo(
+        const sp<MetaData> &meta, const void *asi, size_t asiSize) {
+    static const uint8_t kStaticESDS[] = {
+        0x03, 22,
+        0x00, 0x00,     // ES_ID
+        0x00,           // streamDependenceFlag, URL_Flag, OCRstreamFlag
+
+        0x04, 17,
+        0x40,                       // Audio ISO/IEC 14496-3
+        0x00, 0x00, 0x00, 0x00,
+        0x00, 0x00, 0x00, 0x00,
+        0x00, 0x00, 0x00, 0x00,
+
+        0x05,
+        // AudioSpecificInfo (with size prefix) follows
+    };
+
+    CHECK(asiSize < 128);
+    size_t esdsSize = sizeof(kStaticESDS) + asiSize + 1;
+    uint8_t *esds = new uint8_t[esdsSize];
+    memcpy(esds, kStaticESDS, sizeof(kStaticESDS));
+    uint8_t *ptr = esds + sizeof(kStaticESDS);
+    *ptr++ = asiSize;
+    memcpy(ptr, asi, asiSize);
+
+    meta->setData(kKeyESDS, 0, esds, esdsSize);
+
+    delete[] esds;
+    esds = NULL;
+}
+
+void addVorbisCodecInfo(
+        const sp<MetaData> &meta,
+        const void *_codecPrivate, size_t codecPrivateSize) {
+    // printf("vorbis private data follows:\n");
+    // hexdump(_codecPrivate, codecPrivateSize);
+
+    CHECK(codecPrivateSize >= 3);
+
+    const uint8_t *codecPrivate = (const uint8_t *)_codecPrivate;
+    CHECK(codecPrivate[0] == 0x02);
+
+    size_t len1 = codecPrivate[1];
+    size_t len2 = codecPrivate[2];
+
+    CHECK(codecPrivateSize > 3 + len1 + len2);
+
+    CHECK(codecPrivate[3] == 0x01);
+    meta->setData(kKeyVorbisInfo, 0, &codecPrivate[3], len1);
+
+    CHECK(codecPrivate[len1 + 3] == 0x03);
+
+    CHECK(codecPrivate[len1 + len2 + 3] == 0x05);
+    meta->setData(
+            kKeyVorbisBooks, 0, &codecPrivate[len1 + len2 + 3],
+            codecPrivateSize - len1 - len2 - 3);
+}
+
+void MatroskaExtractor::addTracks() {
+    const mkvparser::Tracks *tracks = mSegment->GetTracks();
+
+    for (size_t index = 0; index < tracks->GetTracksCount(); ++index) {
+        const mkvparser::Track *track = tracks->GetTrackByIndex(index);
+
+        const char *const codecID = track->GetCodecId();
+        LOGV("codec id = %s", codecID);
+        LOGV("codec name = %s", track->GetCodecNameAsUTF8());
+
+        size_t codecPrivateSize;
+        const unsigned char *codecPrivate =
+            track->GetCodecPrivate(&codecPrivateSize);
+
+        enum { VIDEO_TRACK = 1, AUDIO_TRACK = 2 };
+
+        sp<MetaData> meta = new MetaData;
+
+        switch (track->GetType()) {
+            case VIDEO_TRACK:
+            {
+                const mkvparser::VideoTrack *vtrack =
+                    static_cast<const mkvparser::VideoTrack *>(track);
+
+                if (!strcmp("V_MPEG4/ISO/AVC", codecID)) {
+                    meta->setCString(kKeyMIMEType, MEDIA_MIMETYPE_VIDEO_AVC);
+                    meta->setData(kKeyAVCC, 0, codecPrivate, codecPrivateSize);
+                } else if (!strcmp("V_VP8", codecID)) {
+                    meta->setCString(kKeyMIMEType, MEDIA_MIMETYPE_VIDEO_VPX);
+                } else {
+                    continue;
+                }
+
+                meta->setInt32(kKeyWidth, vtrack->GetWidth());
+                meta->setInt32(kKeyHeight, vtrack->GetHeight());
+                break;
+            }
+
+            case AUDIO_TRACK:
+            {
+                const mkvparser::AudioTrack *atrack =
+                    static_cast<const mkvparser::AudioTrack *>(track);
+
+                if (!strcmp("A_AAC", codecID)) {
+                    meta->setCString(kKeyMIMEType, MEDIA_MIMETYPE_AUDIO_AAC);
+                    CHECK(codecPrivateSize >= 2);
+
+                    addESDSFromAudioSpecificInfo(
+                            meta, codecPrivate, codecPrivateSize);
+                } else if (!strcmp("A_VORBIS", codecID)) {
+                    meta->setCString(kKeyMIMEType, MEDIA_MIMETYPE_AUDIO_VORBIS);
+
+                    addVorbisCodecInfo(meta, codecPrivate, codecPrivateSize);
+                } else {
+                    continue;
+                }
+
+                meta->setInt32(kKeySampleRate, atrack->GetSamplingRate());
+                meta->setInt32(kKeyChannelCount, atrack->GetChannels());
+                break;
+            }
+
+            default:
+                continue;
+        }
+
+        long long durationNs = mSegment->GetDuration();
+        meta->setInt64(kKeyDuration, (durationNs + 500) / 1000);
+
+        mTracks.push();
+        TrackInfo *trackInfo = &mTracks.editItemAt(mTracks.size() - 1);
+        trackInfo->mTrackNum = track->GetNumber();
+        trackInfo->mMeta = meta;
+    }
+}
+
+sp<MetaData> MatroskaExtractor::getMetaData() {
+    sp<MetaData> meta = new MetaData;
+    meta->setCString(kKeyMIMEType, MEDIA_MIMETYPE_CONTAINER_MATROSKA);
+
+    return meta;
+}
+
+bool SniffMatroska(
+        const sp<DataSource> &source, String8 *mimeType, float *confidence) {
+    DataSourceReader reader(source);
+    mkvparser::EBMLHeader ebmlHeader;
+    long long pos;
+    if (ebmlHeader.Parse(&reader, pos) < 0) {
+        return false;
+    }
+
+    mimeType->setTo(MEDIA_MIMETYPE_CONTAINER_MATROSKA);
+    *confidence = 0.6;
+
+    return true;
+}
+
+}  // namespace android
diff --git a/media/libstagefright/matroska/MatroskaExtractor.h b/media/libstagefright/matroska/MatroskaExtractor.h
new file mode 100644
index 0000000..7bf41a9
--- /dev/null
+++ b/media/libstagefright/matroska/MatroskaExtractor.h
@@ -0,0 +1,74 @@
+/*
+ * Copyright (C) 2010 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#ifndef MATROSKA_EXTRACTOR_H_
+
+#define MATROSKA_EXTRACTOR_H_
+
+#include <media/stagefright/MediaExtractor.h>
+#include <utils/Vector.h>
+
+namespace mkvparser {
+struct Segment;
+};
+
+namespace android {
+
+class String8;
+
+struct DataSourceReader;
+struct MatroskaSource;
+
+struct MatroskaExtractor : public MediaExtractor {
+    MatroskaExtractor(const sp<DataSource> &source);
+
+    virtual size_t countTracks();
+
+    virtual sp<MediaSource> getTrack(size_t index);
+
+    virtual sp<MetaData> getTrackMetaData(
+            size_t index, uint32_t flags);
+
+    virtual sp<MetaData> getMetaData();
+
+protected:
+    virtual ~MatroskaExtractor();
+
+private:
+    friend struct MatroskaSource;
+
+    struct TrackInfo {
+        unsigned long mTrackNum;
+        sp<MetaData> mMeta;
+    };
+    Vector<TrackInfo> mTracks;
+
+    sp<DataSource> mDataSource;
+    DataSourceReader *mReader;
+    mkvparser::Segment *mSegment;
+
+    void addTracks();
+
+    MatroskaExtractor(const MatroskaExtractor &);
+    MatroskaExtractor &operator=(const MatroskaExtractor &);
+};
+
+bool SniffMatroska(
+        const sp<DataSource> &source, String8 *mimeType, float *confidence);
+
+}  // namespace android
+
+#endif  // MATROSKA_EXTRACTOR_H_
diff --git a/media/libstagefright/matroska/mkvparser.cpp b/media/libstagefright/matroska/mkvparser.cpp
new file mode 100644
index 0000000..4e51004
--- /dev/null
+++ b/media/libstagefright/matroska/mkvparser.cpp
@@ -0,0 +1,3103 @@
+#include "mkvparser.hpp"
+#include <cassert>
+#include <cstring>
+
+mkvparser::IMkvReader::~IMkvReader()
+{
+}
+
+long long mkvparser::ReadUInt(IMkvReader* pReader, long long pos, long& len)
+{
+    assert(pReader);
+    assert(pos >= 0);
+    
+    long long total, available;
+
+    long hr = pReader->Length(&total, &available);
+    assert(hr >= 0);
+    assert(pos < available);
+    assert((available - pos) >= 1);  //assume here max u-int len is 8
+    
+    unsigned char b;
+
+    hr = pReader->Read(pos, 1, &b);
+    if (hr < 0)
+        return hr;
+        
+    assert(hr == 0L);
+    
+    if (b & 0x80)       //1000 0000
+    {
+        len = 1;
+        b &= 0x7F;      //0111 1111
+    }        
+    else if (b & 0x40)  //0100 0000
+    {
+        len = 2;
+        b &= 0x3F;      //0011 1111
+    }
+    else if (b & 0x20)  //0010 0000
+    {
+        len = 3;
+        b &= 0x1F;      //0001 1111
+    }
+    else if (b & 0x10)  //0001 0000
+    {
+        len = 4;
+        b &= 0x0F;      //0000 1111
+    }
+    else if (b & 0x08)  //0000 1000
+    {
+        len = 5;
+        b &= 0x07;      //0000 0111
+    }
+    else if (b & 0x04)  //0000 0100
+    {
+        len = 6;
+        b &= 0x03;      //0000 0011
+    }
+    else if (b & 0x02)  //0000 0010
+    {
+        len = 7;
+        b &= 0x01;      //0000 0001
+    }
+    else 
+    {
+        assert(b & 0x01);  //0000 0001
+        len = 8;
+        b = 0;             //0000 0000
+    }
+    
+    assert((available - pos) >= len);
+    
+    long long result = b;
+    ++pos;
+    for (long i = 1; i < len; ++i)
+    {
+        hr = pReader->Read(pos, 1, &b);
+        
+        if (hr < 0)
+            return hr;
+            
+        assert(hr == 0L);
+        
+        result <<= 8;
+        result |= b;
+        
+        ++pos;
+    }
+    
+    return result;
+}
+    
+    
+long long mkvparser::GetUIntLength(
+    IMkvReader* pReader,
+    long long pos, 
+    long& len)
+{
+    assert(pReader);
+    assert(pos >= 0);
+    
+    long long total, available;
+
+    long hr = pReader->Length(&total, &available);
+    assert(hr >= 0);
+    assert(available <= total);
+    
+    if (pos >= available)
+        return pos;  //too few bytes available
+    
+    unsigned char b;
+    
+    hr = pReader->Read(pos, 1, &b);
+    
+    if (hr < 0)
+        return hr;
+
+    assert(hr == 0L);
+    
+    if (b == 0)  //we can't handle u-int values larger than 8 bytes
+        return E_FILE_FORMAT_INVALID;
+    
+    unsigned char m = 0x80;
+    len = 1;
+    
+    while (!(b & m))
+    {
+        m >>= 1;
+        ++len;
+    }
+    
+    return 0;  //success
+}
+
+
+long long mkvparser::SyncReadUInt(
+    IMkvReader* pReader,
+    long long pos, 
+    long long stop,
+    long& len)
+{
+    assert(pReader);
+
+    if (pos >= stop)
+        return E_FILE_FORMAT_INVALID;
+    
+    unsigned char b;
+    
+    long hr = pReader->Read(pos, 1, &b);
+    
+    if (hr < 0)
+        return hr;
+        
+    if (hr != 0L)
+        return E_BUFFER_NOT_FULL;
+
+    if (b == 0)  //we can't handle u-int values larger than 8 bytes
+        return E_FILE_FORMAT_INVALID;
+    
+    unsigned char m = 0x80;
+    len = 1;
+        
+    while (!(b & m))
+    {
+        m >>= 1;
+        ++len;
+    }
+    
+    if ((pos + len) > stop)
+        return E_FILE_FORMAT_INVALID;
+        
+    long long result = b & (~m);
+    ++pos;
+    
+    for (int i = 1; i < len; ++i)
+    {
+        hr = pReader->Read(pos, 1, &b);
+        
+        if (hr < 0)
+            return hr;
+           
+        if (hr != 0L)
+            return E_BUFFER_NOT_FULL;
+            
+        result <<= 8;
+        result |= b;
+        
+        ++pos;
+    }
+    
+    return result;
+}
+
+
+long long mkvparser::UnserializeUInt(
+    IMkvReader* pReader, 
+    long long pos,
+    long long size)
+{
+    assert(pReader);
+    assert(pos >= 0);
+    assert(size > 0);
+    assert(size <= 8);
+    
+    long long result = 0;
+    
+    for (long long i = 0; i < size; ++i)
+    {
+        unsigned char b;
+        
+        const long hr = pReader->Read(pos, 1, &b);
+        
+        if (hr < 0)      
+            return hr;
+        result <<= 8;
+        result |= b;
+        
+        ++pos;
+    }
+    
+    return result;
+}
+
+
+float mkvparser::Unserialize4Float(
+    IMkvReader* pReader, 
+    long long pos)
+{
+    assert(pReader);
+    assert(pos >= 0);
+    
+    long long total, available;
+    
+    long hr = pReader->Length(&total, &available);
+    assert(hr >= 0);
+    assert(available <= total);
+    assert((pos + 4) <= available);
+    
+    float result;
+    
+    unsigned char* const p = (unsigned char*)&result;
+    unsigned char* q = p + 4;
+    
+    for (;;)
+    {
+        hr = pReader->Read(pos, 1, --q);
+        assert(hr == 0L);
+        
+        if (q == p)
+            break;
+            
+        ++pos;
+    }
+    
+    return result;
+}
+
+
+double mkvparser::Unserialize8Double(
+    IMkvReader* pReader, 
+    long long pos)
+{
+    assert(pReader);
+    assert(pos >= 0);
+    
+    double result;
+    
+    unsigned char* const p = (unsigned char*)&result;
+    unsigned char* q = p + 8;
+    
+    for (;;)
+    {
+        const long hr = pReader->Read(pos, 1, --q);
+        assert(hr == 0L);
+        
+        if (q == p)
+            break;
+            
+        ++pos;
+    }
+    
+    return result;
+}
+
+signed char mkvparser::Unserialize1SInt(
+    IMkvReader* pReader,
+    long long pos)
+{
+    assert(pReader);
+    assert(pos >= 0);
+ 
+    long long total, available;
+
+    long hr = pReader->Length(&total, &available);
+    assert(hr == 0);   
+    assert(available <= total);
+    assert(pos < available);
+
+    signed char result;
+
+    hr = pReader->Read(pos, 1, (unsigned char*)&result);
+    assert(hr == 0);
+
+    return result;
+}
+
+short mkvparser::Unserialize2SInt(
+    IMkvReader* pReader, 
+    long long pos)
+{
+    assert(pReader);
+    assert(pos >= 0);
+    
+    long long total, available;
+    
+    long hr = pReader->Length(&total, &available);
+    assert(hr >= 0);
+    assert(available <= total);
+    assert((pos + 2) <= available);
+    
+    short result;
+    
+    unsigned char* const p = (unsigned char*)&result;
+    unsigned char* q = p + 2;
+    
+    for (;;)
+    {
+        hr = pReader->Read(pos, 1, --q);
+        assert(hr == 0L);
+        
+        if (q == p)
+            break;
+            
+        ++pos;
+    }
+    
+    return result;
+}
+
+
+bool mkvparser::Match(
+    IMkvReader* pReader,
+    long long& pos,
+    unsigned long id_,
+    long long& val)
+
+{
+    assert(pReader);
+    assert(pos >= 0);
+    
+    long long total, available;
+
+    long hr = pReader->Length(&total, &available);
+    assert(hr >= 0);
+    assert(available <= total);
+    
+    long len;
+
+    const long long id = ReadUInt(pReader, pos, len);
+    assert(id >= 0);
+    assert(len > 0);
+    assert(len <= 8);
+    assert((pos + len) <= available);
+    
+    if ((unsigned long)id != id_)
+        return false;
+        
+    pos += len;  //consume id
+    
+    const long long size = ReadUInt(pReader, pos, len);
+    assert(size >= 0);
+    assert(size <= 8);
+    assert(len > 0);
+    assert(len <= 8);
+    assert((pos + len) <= available);
+    
+    pos += len;  //consume length of size of payload
+    
+    val = UnserializeUInt(pReader, pos, size);
+    assert(val >= 0);
+    
+    pos += size;  //consume size of payload
+    
+    return true;
+}
+
+bool mkvparser::Match(
+    IMkvReader* pReader,
+    long long& pos,
+    unsigned long id_,
+    char*& val)
+{
+    assert(pReader);
+    assert(pos >= 0);
+    
+    long long total, available;
+
+    long hr = pReader->Length(&total, &available);
+    assert(hr >= 0);
+    assert(available <= total);
+    
+    long len;
+
+    const long long id = ReadUInt(pReader, pos, len);
+    assert(id >= 0);
+    assert(len > 0);
+    assert(len <= 8);
+    assert((pos + len) <= available);
+    
+    if ((unsigned long)id != id_)
+        return false;
+    
+    pos += len;  //consume id
+    
+    const long long size_ = ReadUInt(pReader, pos, len);
+    assert(size_ >= 0);
+    assert(len > 0);
+    assert(len <= 8);
+    assert((pos + len) <= available);
+    
+    pos += len;  //consume length of size of payload
+    assert((pos + size_) <= available);
+
+    const size_t size = static_cast<size_t>(size_);    
+    val = new char[size+1];
+
+    for (size_t i = 0; i < size; ++i)
+    {
+        char c;
+
+        hr = pReader->Read(pos + i, 1, (unsigned char*)&c);
+        assert(hr == 0L);
+            
+        val[i] = c;
+   
+        if (c == '\0')
+            break;     
+   
+    }
+
+    val[size] = '\0';
+    pos += size_;  //consume size of payload
+    
+    return true;
+}
+
+#if 0
+bool mkvparser::Match(
+    IMkvReader* pReader,
+    long long& pos,
+    unsigned long id,
+    wchar_t*& val)
+{
+    char* str;
+    
+    if (!Match(pReader, pos, id, str))
+        return false;
+
+    const size_t size = mbstowcs(NULL, str, 0);
+       
+    if (size == 0) 
+        val = NULL;
+    else 
+    { 
+        val = new wchar_t[size+1];
+        mbstowcs(val, str, size);
+        val[size] = L'\0';
+    }
+
+    delete[] str;
+    return true;    
+}
+#endif
+
+
+bool mkvparser::Match(
+    IMkvReader* pReader,
+    long long& pos,
+    unsigned long id_,
+    unsigned char*& val,
+    size_t *optionalSize)
+{
+    assert(pReader);
+    assert(pos >= 0);
+    
+    long long total, available;
+
+    long hr = pReader->Length(&total, &available);
+    assert(hr >= 0);
+    assert(available <= total);
+    
+    long len;
+    const long long id = ReadUInt(pReader, pos, len);
+    assert(id >= 0);
+    assert(len > 0);
+    assert(len <= 8);
+    assert((pos + len) <= available);
+    
+    if ((unsigned long)id != id_)
+        return false;
+        
+    pos += len;  //consume id
+    
+    const long long size_ = ReadUInt(pReader, pos, len);
+    assert(size_ >= 0);
+    assert(len > 0);
+    assert(len <= 8);
+    assert((pos + len) <= available);
+    
+    pos += len;  //consume length of size of payload
+    assert((pos + size_) <= available);
+
+    const size_t size = static_cast<size_t>(size_);    
+    val = new unsigned char[size];
+ 
+    if (optionalSize) {
+        *optionalSize = size;
+    }
+
+    for (size_t i = 0; i < size; ++i)
+    {
+        unsigned char b;
+
+        hr = pReader->Read(pos + i, 1, &b);
+        assert(hr == 0L);
+
+        val[i] = b; 
+    }
+    
+    pos += size_;  //consume size of payload    
+    return true;
+}
+
+
+bool mkvparser::Match(
+    IMkvReader* pReader,
+    long long& pos,
+    unsigned long id_,
+    double& val)
+{
+    assert(pReader);
+    assert(pos >= 0);
+    
+    long long total, available;
+
+    long hr = pReader->Length(&total, &available);
+    assert(hr >= 0);
+    assert(available <= total);
+    long idlen;
+    const long long id = ReadUInt(pReader, pos, idlen);
+    assert(id >= 0);  //TODO
+    
+    if ((unsigned long)id != id_)
+        return false;
+
+    long sizelen;
+    const long long size = ReadUInt(pReader, pos + idlen, sizelen);
+
+    switch (size)
+    {	
+        case 4:
+        case 8:
+            break;
+        default:
+            return false;
+    }
+
+    pos += idlen + sizelen;  //consume id and size fields
+    assert((pos + size) <= available);
+
+    if (size == 4)
+        val = Unserialize4Float(pReader, pos);
+    else
+    {
+        assert(size == 8);
+        val = Unserialize8Double(pReader, pos);
+    }
+    
+    pos += size;  //consume size of payload
+    
+    return true;
+}
+
+
+bool mkvparser::Match(
+    IMkvReader* pReader,
+    long long& pos,
+    unsigned long id_,
+    short& val)
+{
+    assert(pReader);
+    assert(pos >= 0);
+    
+    long long total, available;
+
+    long hr = pReader->Length(&total, &available);
+    assert(hr >= 0);
+    assert(available <= total);
+    
+    long len;
+    const long long id = ReadUInt(pReader, pos, len);
+    assert(id >= 0);
+    assert((pos + len) <= available);
+    
+    if ((unsigned long)id != id_)
+        return false;
+        
+    pos += len;  //consume id
+    
+    const long long size = ReadUInt(pReader, pos, len);
+    assert(size <= 2);
+    assert((pos + len) <= available);
+   
+    pos += len;  //consume length of size of payload
+    assert((pos + size) <= available);
+
+    //TODO:
+    // Generalize this to work for any size signed int
+    if (size == 1)
+        val = Unserialize1SInt(pReader, pos);
+    else 
+        val = Unserialize2SInt(pReader, pos);
+        
+    pos += size;  //consume size of payload
+    
+    return true;
+}
+
+
+namespace mkvparser
+{
+
+EBMLHeader::EBMLHeader():
+    m_docType(NULL)
+{
+}
+
+EBMLHeader::~EBMLHeader()
+{
+    delete[] m_docType;
+}
+
+long long EBMLHeader::Parse(
+    IMkvReader* pReader,
+    long long& pos)
+{
+    assert(pReader);
+    
+    long long total, available;
+    
+    long hr = pReader->Length(&total, &available);
+    
+    if (hr < 0) 
+        return hr;
+    
+    pos = 0;    
+    long long end = (1024 < available)? 1024: available;    
+
+    for (;;)
+    {    
+        unsigned char b = 0;
+    
+        while (pos < end)
+        {
+            hr = pReader->Read(pos, 1, &b);
+           
+            if (hr < 0)
+                return hr;
+            
+            if (b == 0x1A)
+                break;
+                
+            ++pos;
+        }
+    
+        if (b != 0x1A)
+        {
+            if ((pos >= 1024) ||
+                (available >= total) || 
+                ((total - available) < 5))
+                  return -1;
+                
+            return available + 5;  //5 = 4-byte ID + 1st byte of size
+        }
+    
+        if ((total - pos) < 5)
+            return E_FILE_FORMAT_INVALID;
+            
+        if ((available - pos) < 5)
+            return pos + 5;  //try again later
+
+        long len;            
+
+        const long long result = ReadUInt(pReader, pos, len);
+        
+        if (result < 0)  //error
+            return result;
+            
+        if (result == 0x0A45DFA3)  //ReadId masks-off length indicator bits
+        {
+            assert(len == 4);
+            pos += len;
+            break;
+        }
+
+        ++pos;  //throw away just the 0x1A byte, and try again
+    }
+        
+    long len;
+    long long result = GetUIntLength(pReader, pos, len);
+    
+    if (result < 0)  //error
+        return result;
+        
+    if (result > 0)  //need more data
+        return result;
+        
+    assert(len > 0);
+    assert(len <= 8);
+    
+    if ((total -  pos) < len)
+        return E_FILE_FORMAT_INVALID;
+    if ((available - pos) < len)
+        return pos + len;  //try again later
+        
+    result = ReadUInt(pReader, pos, len);
+    
+    if (result < 0)  //error
+        return result;
+        
+    pos += len;  //consume u-int
+    
+    if ((total - pos) < result)
+        return E_FILE_FORMAT_INVALID;
+
+    if ((available - pos) < result)
+        return pos + result;
+        
+    end = pos + result;
+    
+    m_version = 1;
+    m_readVersion = 1;
+    m_maxIdLength = 4;
+    m_maxSizeLength = 8;
+    m_docTypeVersion = 1;
+    m_docTypeReadVersion = 1;
+
+    while (pos < end)
+    {
+        if (Match(pReader, pos, 0x0286, m_version))   
+            ;
+        else if (Match(pReader, pos, 0x02F7, m_readVersion))        
+            ;
+        else if (Match(pReader, pos, 0x02F2, m_maxIdLength))        
+            ;
+        else if (Match(pReader, pos, 0x02F3, m_maxSizeLength))      
+            ;
+        else if (Match(pReader, pos, 0x0282, m_docType))            
+            ; 
+        else if (Match(pReader, pos, 0x0287, m_docTypeVersion))     
+            ;
+        else if (Match(pReader, pos, 0x0285, m_docTypeReadVersion)) 
+            ;
+        else
+        {
+            result = ReadUInt(pReader, pos, len);
+            assert(result > 0);
+            assert(len > 0);
+            assert(len <= 8);
+        
+            pos += len;
+            assert(pos < end);
+            
+            result = ReadUInt(pReader, pos, len);
+            assert(result >= 0);
+            assert(len > 0);
+            assert(len <= 8);
+            
+            pos += len + result;
+            assert(pos <= end);
+        }
+    }
+    
+    assert(pos == end);
+        
+    return 0;    
+}
+
+
+Segment::Segment(
+    IMkvReader* pReader,
+    long long start,
+    long long size) :
+    m_pReader(pReader),
+    m_start(start),
+    m_size(size),
+    m_pos(start),
+    m_pInfo(NULL),
+    m_pTracks(NULL),
+    m_clusterCount(0)
+    //m_clusterNumber(0)
+{
+}
+
+
+Segment::~Segment()
+{
+    Cluster** i = m_clusters;
+    Cluster** j = m_clusters + m_clusterCount;
+
+    while (i != j)
+    {
+        Cluster* p = *i++;
+        assert(p);		
+        delete p;
+    } 
+    
+    delete[] m_clusters;
+       
+    delete m_pTracks;
+    delete m_pInfo;
+}
+
+
+long long Segment::CreateInstance(
+    IMkvReader* pReader,
+    long long pos,
+    Segment*& pSegment)
+{
+    assert(pReader);
+    assert(pos >= 0);
+    
+    pSegment = NULL;
+    
+    long long total, available;
+    
+    long hr = pReader->Length(&total, &available);
+    assert(hr >= 0);
+    assert(available <= total);
+    
+    //I would assume that in practice this loop would execute
+    //exactly once, but we allow for other elements (e.g. Void)
+    //to immediately follow the EBML header.  This is fine for
+    //the source filter case (since the entire file is available),
+    //but in the splitter case over a network we should probably
+    //just give up early.  We could for example decide only to
+    //execute this loop a maximum of, say, 10 times.
+    
+    while (pos < total)
+    {    
+        //Read ID
+        
+        long len;
+        long long result = GetUIntLength(pReader, pos, len);
+        
+        if (result)  //error, or too few available bytes
+            return result;
+            
+        if ((pos + len) > total)
+            return E_FILE_FORMAT_INVALID;
+            
+        if ((pos + len) > available)
+            return pos + len;
+
+        //TODO: if we liberalize the behavior of ReadUInt, we can
+        //probably eliminate having to use GetUIntLength here.
+        const long long id = ReadUInt(pReader, pos, len);
+        
+        if (id < 0)  //error
+            return id;
+            
+        pos += len;  //consume ID
+        
+        //Read Size
+        
+        result = GetUIntLength(pReader, pos, len);
+        
+        if (result)  //error, or too few available bytes
+            return result;
+            
+        if ((pos + len) > total)
+            return E_FILE_FORMAT_INVALID;
+            
+        if ((pos + len) > available)
+            return pos + len;
+
+        //TODO: if we liberalize the behavior of ReadUInt, we can
+        //probably eliminate having to use GetUIntLength here.
+        const long long size = ReadUInt(pReader, pos, len);
+        
+        if (size < 0)
+            return size;
+            
+        pos += len;  //consume length of size of element
+        
+        //Pos now points to start of payload
+        
+        if ((pos + size) > total)
+            return E_FILE_FORMAT_INVALID;
+        
+        if (id == 0x08538067)  //Segment ID
+        {
+            pSegment = new  Segment(pReader, pos, size); 
+            assert(pSegment);  //TODO   
+
+            return 0;    //success
+        }
+        
+        pos += size;  //consume payload
+    }
+    
+    assert(pos == total);
+    
+    pSegment = new Segment(pReader, pos, 0); 
+    assert(pSegment);  //TODO   
+
+    return 0;  //success (sort of)
+}
+
+
+long long Segment::ParseHeaders()
+{
+    //Outermost (level 0) segment object has been constructed, 
+    //and pos designates start of payload.  We need to find the
+    //inner (level 1) elements.
+    long long total, available;
+    
+    long hr = m_pReader->Length(&total, &available);
+    assert(hr >= 0);
+    assert(available <= total);
+    
+    const long long stop = m_start + m_size;
+    assert(stop <= total);
+    assert(m_pos <= stop);
+    
+    bool bQuit = false;
+    while ((m_pos < stop) && !bQuit)
+    {
+        long long pos = m_pos;
+        
+        long len;
+        long long result = GetUIntLength(m_pReader, pos, len);
+        
+        if (result)  //error, or too few available bytes
+            return result;
+            
+        if ((pos + len) > stop)
+            return E_FILE_FORMAT_INVALID;
+            
+        if ((pos + len) > available)
+            return pos + len;
+            
+        const long long idpos = pos;
+        const long long id = ReadUInt(m_pReader, idpos, len);
+        
+        if (id < 0)  //error
+            return id;
+            
+        pos += len;  //consume ID
+        
+        //Read Size
+        result = GetUIntLength(m_pReader, pos, len);
+        
+        if (result)  //error, or too few available bytes
+            return result;
+            
+        if ((pos + len) > stop)
+            return E_FILE_FORMAT_INVALID;
+            
+        if ((pos + len) > available)
+            return pos + len;
+
+        const long long size = ReadUInt(m_pReader, pos, len);
+        
+        if (size < 0)
+            return size;
+            
+        pos += len;  //consume length of size of element
+        
+        //Pos now points to start of payload
+        
+        if ((pos + size) > stop)
+            return E_FILE_FORMAT_INVALID;
+            
+        //We read EBML elements either in total or nothing at all.
+            
+        if ((pos + size) > available)
+            return pos + size;
+        
+        if (id == 0x0549A966)  //Segment Info ID
+        {
+            assert(m_pInfo == NULL);
+            m_pInfo = new  SegmentInfo(this, pos, size);
+            assert(m_pInfo);  //TODO
+            
+            if (m_pTracks)
+                bQuit = true;
+        }
+        else if (id == 0x0654AE6B)  //Tracks ID
+        {
+            assert(m_pTracks == NULL);
+            m_pTracks = new  Tracks(this, pos, size);
+            assert(m_pTracks);  //TODO
+            
+            if (m_pInfo)
+                bQuit = true;
+        }
+        else if (id == 0x0F43B675)  //Cluster ID
+        {
+#if 0
+            if (m_pInfo == NULL)  //TODO: liberalize
+                ;  
+            else if (m_pTracks == NULL)
+                ;
+            else
+                //ParseCluster(idpos, pos, size);            
+                Cluster::Parse(this, m_clusters, pos, size);
+#endif
+            bQuit = true;
+        }
+        
+        m_pos = pos + size;  //consume payload
+    }
+    
+    assert(m_pos <= stop);
+    
+    return 0;  //success
+}
+
+
+long Segment::ParseCluster(Cluster*& pCluster, long long& pos_) const
+{
+    pCluster = NULL;
+    pos_ = -1;
+    
+    const long long stop = m_start + m_size;
+    assert(m_pos <= stop);
+    
+    long long pos = m_pos;
+    long long off = -1;
+   
+ 
+    while (pos < stop)
+    {
+        long len;
+        const long long idpos = pos;
+        
+        const long long id = SyncReadUInt(m_pReader, pos, stop, len);
+        
+        if (id < 0)  //error
+            return static_cast<long>(id);
+            
+        if (id == 0)
+            return E_FILE_FORMAT_INVALID;
+            
+        pos += len;  //consume id        
+        assert(pos < stop);
+
+        const long long size = SyncReadUInt(m_pReader, pos, stop, len);
+        
+        if (size < 0)  //error
+            return static_cast<long>(size);
+            
+        pos += len;  //consume size
+        assert(pos <= stop);
+            
+        if (size == 0)  //weird
+            continue;
+            
+        //pos now points to start of payload
+            
+        pos += size;  //consume payload
+        assert(pos <= stop);
+
+        if (off >= 0)
+        {
+            pos_ = idpos;
+            break;
+        }
+
+        if (id == 0x0F43B675)  //Cluster ID
+            off = idpos - m_start;
+    }
+    
+    Segment* const this_ = const_cast<Segment*>(this);
+    const size_t idx = m_clusterCount;
+    
+    if (pos >= stop)
+    {
+        pos_ = stop;
+        
+#if 0        
+        if (off < 0)
+        {
+            pCluster = Cluster::CreateEndOfStream(this_, idx);
+            return 1L;
+        }
+#else
+        if (off < 0)
+            return 1L;
+#endif
+                
+        //Reading 0 bytes at pos might work too -- it would depend 
+        //on how the reader is implemented.
+        
+        unsigned char b;
+
+        const long hr = m_pReader->Read(pos - 1, 1, &b);
+        
+        if (hr < 0)
+            return hr;
+            
+        if (hr != 0L)
+            return E_BUFFER_NOT_FULL;
+    }
+    
+    assert(off >= 0);
+    assert(pos_ >= m_start);
+    assert(pos_ <= stop);
+
+    pCluster = Cluster::Parse(this_, idx, off);
+    return 0L;
+}
+
+
+bool Segment::AddCluster(Cluster* pCluster, long long pos)
+{
+    assert(pos >= m_start);
+    
+    const long long stop = m_start + m_size;
+    assert(pos <= stop);
+
+    if (pCluster)    
+        m_clusters[pos] = pCluster;
+        
+    m_pos = pos;  //m_pos >= stop is now we know we have all clusters
+    
+    return (pos >= stop);
+}
+
+
+long Segment::Load()
+{
+    //Outermost (level 0) segment object has been constructed, 
+    //and pos designates start of payload.  We need to find the
+    //inner (level 1) elements.
+    const long long stop = m_start + m_size;
+#ifdef _DEBUG
+    {
+        long long total, available;
+        
+        long hr = m_pReader->Length(&total, &available);
+        assert(hr >= 0);
+        assert(available >= total);
+        assert(stop <= total);
+    }
+#endif
+    long long index = m_pos;
+    
+    m_clusterCount = 0;
+
+    while (index < stop)
+    {
+        long len = 0;
+
+        long long result = GetUIntLength(m_pReader, index, len);
+       
+        if (result < 0)  //error
+            return static_cast<long>(result);
+            
+        if ((index + len) > stop)
+            return E_FILE_FORMAT_INVALID;
+            
+        const long long idpos = index;
+        const long long id = ReadUInt(m_pReader, idpos, len);
+        
+        if (id < 0)  //error
+            return static_cast<long>(id);
+            
+        index += len;  //consume ID
+        
+        //Read Size
+        result = GetUIntLength(m_pReader, index, len);
+        
+        if (result < 0)  //error
+            return static_cast<long>(result);
+            
+        if ((index + len) > stop)
+            return E_FILE_FORMAT_INVALID;
+            
+        const long long size = ReadUInt(m_pReader, index, len);
+        
+        if (size < 0)  //error
+            return static_cast<long>(size);
+            
+        index += len;  //consume length of size of element
+ 
+        if (id == 0x0F43B675) // Cluster ID 
+            break;
+	
+        if (id == 0x014D9B74) // SeekHead ID 
+        {
+            ParseSeekHead(index, size, NULL); 
+            break;
+        }
+        index += size;
+    }
+        
+    if (m_clusterCount == 0)
+        return -1L;
+
+    while (m_pos < stop)
+    {
+        long long pos = m_pos;
+        
+        long len;
+
+        long long result = GetUIntLength(m_pReader, pos, len);
+        
+        if (result < 0)  //error
+            return static_cast<long>(result);
+            
+        if ((pos + len) > stop)
+            return E_FILE_FORMAT_INVALID;
+            
+        const long long idpos = pos;
+        const long long id = ReadUInt(m_pReader, idpos, len);
+        
+        if (id < 0)  //error
+            return static_cast<long>(id);
+            
+        pos += len;  //consume ID
+        
+        //Read Size
+        result = GetUIntLength(m_pReader, pos, len);
+        
+        if (result < 0)  //error
+            return static_cast<long>(result);
+            
+        if ((pos + len) > stop)
+	        return E_FILE_FORMAT_INVALID;
+            
+        const long long size = ReadUInt(m_pReader, pos, len);
+       
+        if (size < 0)  //error
+            return static_cast<long>(size);
+            
+        pos += len;  //consume length of size of element
+        
+        //Pos now points to start of payload
+        
+        if ((pos + size) > stop)
+            return E_FILE_FORMAT_INVALID;
+            
+        if (id == 0x0F43B675)  //Cluster ID
+            break;
+
+        if (id == 0x014D9B74)  //SeekHead ID
+        {
+            m_clusters = new Cluster*[m_clusterCount];   
+            size_t index = 0;
+            
+            ParseSeekHead(pos, size, &index);            
+            assert(index == m_clusterCount);
+        }            
+        else if (id == 0x0549A966)  //Segment Info ID
+        {
+            assert(m_pInfo == NULL);
+            m_pInfo = new  SegmentInfo(this, pos, size);
+            assert(m_pInfo);  //TODO
+        }
+        else if (id == 0x0654AE6B)  //Tracks ID
+        {
+            assert(m_pTracks == NULL);
+            m_pTracks = new Tracks(this, pos, size);
+            assert(m_pTracks);  //TODO
+        }
+
+        m_pos = pos + size;  //consume payload
+    }
+    
+    assert(m_clusters);
+    
+    //TODO: see notes above.  This check is here (temporarily) to ensure
+    //that the first seekhead has entries for the clusters (because that's
+    //when they're loaded).  In case we are given a file that lists the
+    //clusters in a second seekhead, the worst thing that happens is that
+    //we treat this as an invalid file (which is better then simply
+    //asserting somewhere).  But that's only a work-around.  What we need
+    //to do is be able to handle having multiple seekheads, and having
+    //clusters listed somewhere besides the first seekhead.
+    //    
+    //if (m_clusters == NULL)
+    //    return E_FILE_FORMAT_INVALID;
+        
+    //NOTE: we stop parsing when we reach the first cluster, under the
+    //assumption all clusters are named in some SeekHead.  Clusters
+    //will have been (pre)loaded, so we indicate that we have all clusters
+    //by adjusting the parse position:
+    m_pos = stop;  //means "we have all clusters"
+
+    return 0L;
+}
+
+
+void Segment::ParseSeekHead(long long start, long long size_, size_t* pIndex)
+{
+    long long pos = start;
+    const long long stop = start + size_;
+    while (pos < stop)
+    {
+        long len;
+        
+        const long long id = ReadUInt(m_pReader, pos, len);
+        assert(id >= 0);  //TODO
+        assert((pos + len) <= stop);
+        
+        pos += len;  //consume ID
+        
+        const long long size = ReadUInt(m_pReader, pos, len);
+        assert(size >= 0);
+        assert((pos + len) <= stop);
+        
+        pos += len;  //consume Size field
+        assert((pos + size) <= stop);
+
+        if (id == 0x0DBB)  //SeekEntry ID
+            ParseSeekEntry(pos, size, pIndex);
+        
+        pos += size;  //consume payload
+        assert(pos <= stop);
+    }
+    
+    assert(pos == stop);
+}
+
+
+void Segment::ParseSecondarySeekHead(long long off, size_t* pIndex)
+{
+    assert(off >= 0);
+    assert(off < m_size);
+
+    long long pos = m_start + off;
+    const long long stop = m_start + m_size;
+    
+    long len;
+
+    long long result = GetUIntLength(m_pReader, pos, len);
+    assert(result == 0);
+    assert((pos + len) <= stop);
+    
+    const long long idpos = pos;
+
+    const long long id = ReadUInt(m_pReader, idpos, len);
+    assert(id == 0x014D9B74);  //SeekHead ID
+    
+    pos += len;  //consume ID
+    assert(pos < stop);
+    
+    //Read Size
+    
+    result = GetUIntLength(m_pReader, pos, len);
+    assert(result == 0);
+    assert((pos + len) <= stop);
+    
+    const long long size = ReadUInt(m_pReader, pos, len);
+    assert(size >= 0);
+    
+    pos += len;  //consume length of size of element
+    assert((pos + size) <= stop);
+    
+    //Pos now points to start of payload
+    
+    ParseSeekHead(pos, size, pIndex);
+}
+
+
+void Segment::ParseSeekEntry(long long start, long long size_, size_t* pIndex)
+{
+    long long pos = start;
+
+    const long long stop = start + size_;
+    
+    long len;
+    
+    const long long seekIdId = ReadUInt(m_pReader, pos, len);
+    //seekIdId;
+    assert(seekIdId == 0x13AB);  //SeekID ID
+    assert((pos + len) <= stop);
+    
+    pos += len;  //consume id
+
+    const long long seekIdSize = ReadUInt(m_pReader, pos, len);
+    assert(seekIdSize >= 0);
+    assert((pos + len) <= stop);
+    
+    pos += len;  //consume size
+    
+    const long long seekId = ReadUInt(m_pReader, pos, len);  //payload
+    assert(seekId >= 0);
+    assert(len == seekIdSize);
+    assert((pos + len) <= stop);
+    
+    pos += seekIdSize;  //consume payload
+    
+    const long long seekPosId = ReadUInt(m_pReader, pos, len);
+    //seekPosId;
+    assert(seekPosId == 0x13AC);  //SeekPos ID
+    assert((pos + len) <= stop);
+    
+    pos += len;  //consume id
+    
+    const long long seekPosSize = ReadUInt(m_pReader, pos, len);
+    assert(seekPosSize >= 0);
+    assert((pos + len) <= stop);
+
+    pos += len;  //consume size
+    assert((pos + seekPosSize) <= stop);
+        
+    const long long seekOff = UnserializeUInt(m_pReader, pos, seekPosSize);
+    assert(seekOff >= 0);
+    assert(seekOff < m_size);
+    
+    pos += seekPosSize;  //consume payload
+    assert(pos == stop);
+    
+    const long long seekPos = m_start + seekOff;
+    assert(seekPos < (m_start + m_size));
+   
+    if (seekId == 0x0F43B675)  //cluster id
+    {       
+        if (pIndex == NULL)
+            ++m_clusterCount; 
+        else
+        {
+            assert(m_clusters);
+            assert(m_clusterCount > 0);
+            
+            size_t& index = *pIndex;
+            assert(index < m_clusterCount);
+            
+            Cluster*& pCluster = m_clusters[index];
+            
+            pCluster = Cluster::Parse(this, index, seekOff);
+            assert(pCluster);  //TODO
+            
+            ++index;
+        }
+    }
+    else if (seekId == 0x014D9B74)  //SeekHead ID
+    {
+        ParseSecondarySeekHead(seekOff, pIndex);
+    }
+}
+
+
+long long Segment::Unparsed() const
+{
+    const long long stop = m_start + m_size;
+
+    const long long result = stop - m_pos;
+    assert(result >= 0);
+    
+    return result;
+}
+
+
+#if 0  //NOTE: too inefficient
+long long Segment::Load(long long time_ns)
+{
+    if (Unparsed() <= 0)
+        return 0;
+    
+    while (m_clusters.empty())
+    {
+        const long long result = Parse();
+        
+        if (result)  //error, or not enough bytes available
+            return result;
+            
+        if (Unparsed() <= 0)
+            return 0;
+    }
+    
+    while (m_clusters.back()->GetTime() < time_ns)
+    {
+        const long long result = Parse();
+        
+        if (result)  //error, or not enough bytes available
+            return result;
+            
+        if (Unparsed() <= 0)
+            return 0;
+    }        
+
+    return 0;        
+}
+#endif
+
+
+Cluster* Segment::GetFirst()
+{
+    if ((m_clusters == NULL) || (m_clusterCount <= 0))
+       return &m_eos;
+
+    Cluster* const pCluster = m_clusters[0];
+    assert(pCluster);
+        
+    return pCluster;
+}
+
+
+Cluster* Segment::GetLast()
+{
+    if ((m_clusters == NULL) || (m_clusterCount <= 0))
+        return &m_eos;
+
+    const size_t idx = m_clusterCount - 1;    
+    Cluster* const pCluster = m_clusters[idx];
+    assert(pCluster);
+        
+    return pCluster;
+}
+
+
+unsigned long Segment::GetCount() const
+{
+    //TODO: m_clusterCount should not be long long.
+    return static_cast<unsigned long>(m_clusterCount);
+}
+
+
+Cluster* Segment::GetNext(const Cluster* pCurr)
+{
+    assert(pCurr);
+    assert(pCurr != &m_eos);
+    assert(m_clusters);
+    assert(m_clusterCount > 0);
+
+    size_t idx =  pCurr->m_index;
+    assert(idx < m_clusterCount);
+    assert(pCurr == m_clusters[idx]);
+    
+    idx++;
+    
+    if (idx >= m_clusterCount) 
+        return &m_eos;
+        
+    Cluster* const pNext = m_clusters[idx];
+    assert(pNext);
+    
+    return pNext;
+}
+
+
+Cluster* Segment::GetCluster(long long time_ns)
+{
+    if ((m_clusters == NULL) || (m_clusterCount <= 0))
+        return &m_eos;
+        
+    {
+        Cluster* const pCluster = m_clusters[0];
+        assert(pCluster);
+        assert(pCluster->m_index == 0);
+        
+        if (time_ns <= pCluster->GetTime())
+            return pCluster;
+    }
+    
+    //Binary search of cluster array
+       
+    size_t i = 0;
+    size_t j = m_clusterCount;
+    
+    while (i < j)
+    {
+        //INVARIANT:
+        //[0, i) <= time_ns
+        //[i, j) ?
+        //[j, m_clusterCount)  > time_ns
+        
+        const size_t k = i + (j - i) / 2;
+        assert(k < m_clusterCount);
+
+        Cluster* const pCluster = m_clusters[k];
+        assert(pCluster);
+        assert(pCluster->m_index == k);
+        
+        const long long t = pCluster->GetTime();
+        
+        if (t <= time_ns)
+            i = k + 1;
+        else
+            j = k;
+            
+        assert(i <= j);
+    }
+    
+    assert(i == j);
+    assert(i > 0);
+    assert(i <= m_clusterCount);
+    
+    const size_t k = i - 1;
+    
+    Cluster* const pCluster = m_clusters[k];
+    assert(pCluster);
+    assert(pCluster->m_index == k);
+    assert(pCluster->GetTime() <= time_ns);
+    
+    return pCluster;
+}
+
+
+Tracks* Segment::GetTracks() const
+{
+    return m_pTracks;
+}
+
+
+const SegmentInfo* const Segment::GetInfo() const
+{
+    return m_pInfo;
+}
+
+
+long long Segment::GetDuration() const
+{
+    assert(m_pInfo);
+    return m_pInfo->GetDuration();
+}
+
+
+SegmentInfo::SegmentInfo(Segment* pSegment, long long start, long long size_) :
+    m_pSegment(pSegment),
+    m_start(start),
+    m_size(size_),
+    m_pMuxingAppAsUTF8(NULL),
+    m_pWritingAppAsUTF8(NULL),
+    m_pTitleAsUTF8(NULL)
+{
+    IMkvReader* const pReader = m_pSegment->m_pReader;
+   
+    long long pos = start;
+    const long long stop = start + size_;
+    
+    m_timecodeScale = 1000000;
+    m_duration = 0;
+    
+    
+    while (pos < stop)
+    {
+        if (Match(pReader, pos, 0x0AD7B1, m_timecodeScale))
+            assert(m_timecodeScale > 0);
+
+        else if (Match(pReader, pos, 0x0489, m_duration))
+            assert(m_duration >= 0);
+
+        else if (Match(pReader, pos, 0x0D80, m_pMuxingAppAsUTF8))   //[4D][80] 
+            assert(m_pMuxingAppAsUTF8);
+
+        else if (Match(pReader, pos, 0x1741, m_pWritingAppAsUTF8))  //[57][41]
+            assert(m_pWritingAppAsUTF8);
+            
+        else if (Match(pReader, pos, 0x3BA9, m_pTitleAsUTF8))        //[7B][A9]
+            assert(m_pTitleAsUTF8);
+
+        else
+        {
+            long len;
+            
+            const long long id = ReadUInt(pReader, pos, len);
+            //id;
+            assert(id >= 0);
+            assert((pos + len) <= stop);
+            
+            pos += len;  //consume id
+            assert((stop - pos) > 0);
+            
+            const long long size = ReadUInt(pReader, pos, len);
+            assert(size >= 0);
+            assert((pos + len) <= stop);
+            
+            pos += len + size;  //consume size and payload
+            assert(pos <= stop);
+        }
+    }
+    
+    assert(pos == stop);
+}
+
+SegmentInfo::~SegmentInfo()
+{
+    if (m_pMuxingAppAsUTF8)
+    {
+        delete[] m_pMuxingAppAsUTF8;
+        m_pMuxingAppAsUTF8 = NULL;
+    }
+
+    if (m_pWritingAppAsUTF8)
+    {
+        delete[] m_pWritingAppAsUTF8;
+        m_pWritingAppAsUTF8 = NULL;
+    }
+   
+    if (m_pTitleAsUTF8)
+    {
+        delete[] m_pTitleAsUTF8;
+        m_pTitleAsUTF8 = NULL;
+    }
+}
+
+long long SegmentInfo::GetTimeCodeScale() const
+{
+    return m_timecodeScale;
+}
+
+
+long long SegmentInfo::GetDuration() const
+{
+    assert(m_duration >= 0);    
+    assert(m_timecodeScale >= 1);
+    
+    const double dd = double(m_duration) * double(m_timecodeScale);
+    const long long d = static_cast<long long>(dd);
+    
+    return d;
+}
+
+const char* SegmentInfo::GetMuxingAppAsUTF8() const
+{
+    return m_pMuxingAppAsUTF8;
+}
+
+const char* SegmentInfo::GetWritingAppAsUTF8() const
+{
+    return m_pWritingAppAsUTF8;
+}
+
+const char* SegmentInfo::GetTitleAsUTF8() const
+{
+    return m_pTitleAsUTF8;
+}
+
+Track::Track(Segment* pSegment, const Info& i) :
+    m_pSegment(pSegment),
+    m_info(i)
+{
+}
+
+Track::~Track()
+{
+    Info& info = const_cast<Info&>(m_info);
+    info.Clear();
+}
+
+Track::Info::Info():
+    type(-1),
+    number(-1),
+    uid(-1),
+    nameAsUTF8(NULL),
+    codecId(NULL),
+    codecPrivate(NULL),
+    codecPrivateSize(0),
+    codecNameAsUTF8(NULL)
+{
+}
+
+void Track::Info::Clear() 
+{
+    delete[] nameAsUTF8;
+    nameAsUTF8 = NULL;
+
+    delete[] codecId;
+    codecId = NULL;
+
+    delete[] codecPrivate;
+    codecPrivate = NULL;
+
+    delete[] codecNameAsUTF8;
+    codecNameAsUTF8 = NULL;
+}
+
+const BlockEntry* Track::GetEOS() const
+{
+    return &m_eos;
+}
+
+long long Track::GetType() const
+{
+    const unsigned long result = static_cast<unsigned long>(m_info.type);
+    return result;
+}
+
+unsigned long Track::GetNumber() const
+{
+    assert(m_info.number >= 0);
+    const unsigned long result = static_cast<unsigned long>(m_info.number);
+    return result;
+}
+
+const char* Track::GetNameAsUTF8() const
+{
+    return m_info.nameAsUTF8;
+}
+
+const char* Track::GetCodecNameAsUTF8() const
+{  
+    return m_info.codecNameAsUTF8;
+}
+
+
+const char* Track::GetCodecId() const
+{
+    return m_info.codecId;
+}
+
+
+const unsigned char* Track::GetCodecPrivate(size_t *optionalSize) const
+{
+    if (optionalSize) {
+        *optionalSize = m_info.codecPrivateSize;
+    }
+    return m_info.codecPrivate;
+}
+
+
+long Track::GetFirst(const BlockEntry*& pBlockEntry) const
+{
+    Cluster* const pCluster = m_pSegment->GetFirst();
+    
+    //If Segment::GetFirst returns NULL, then this must be a network 
+    //download, and we haven't loaded any clusters yet.  In this case,
+    //returning NULL from Track::GetFirst means the same thing.
+
+    if ((pCluster == NULL) || pCluster->EOS())
+    {
+        pBlockEntry = NULL;
+        return E_BUFFER_NOT_FULL;  //return 1L instead?
+    }
+        
+    pBlockEntry = pCluster->GetFirst();
+    
+    while (pBlockEntry)
+    {
+        const Block* const pBlock = pBlockEntry->GetBlock();
+        assert(pBlock);
+        
+        if (pBlock->GetTrackNumber() == (unsigned long)m_info.number)
+            return 0L;
+            
+        pBlockEntry = pCluster->GetNext(pBlockEntry);
+    }
+    
+    //NOTE: if we get here, it means that we didn't find a block with
+    //a matching track number.  We interpret that as an error (which
+    //might be too conservative).
+
+    pBlockEntry = GetEOS();  //so we can return a non-NULL value
+    return 1L;
+}
+
+
+long Track::GetNext(const BlockEntry* pCurrEntry, const BlockEntry*& pNextEntry) const
+{
+    assert(pCurrEntry);
+    assert(!pCurrEntry->EOS());  //?
+    assert(pCurrEntry->GetBlock()->GetTrackNumber() == (unsigned long)m_info.number);    
+    
+    const Cluster* const pCurrCluster = pCurrEntry->GetCluster();
+    assert(pCurrCluster);
+    assert(!pCurrCluster->EOS());
+    
+    pNextEntry = pCurrCluster->GetNext(pCurrEntry);
+            
+    while (pNextEntry)
+    {    
+        const Block* const pNextBlock = pNextEntry->GetBlock();
+        assert(pNextBlock);
+    
+        if (pNextBlock->GetTrackNumber() == (unsigned long)m_info.number)
+            return 0L;
+            
+        pNextEntry = pCurrCluster->GetNext(pNextEntry);
+    }
+
+    Segment* pSegment = pCurrCluster->m_pSegment;    
+    Cluster* const pNextCluster = pSegment->GetNext(pCurrCluster);
+    
+    if ((pNextCluster == NULL) || pNextCluster->EOS())
+    {
+        if (pSegment->Unparsed() <= 0)   //all clusters have been loaded
+        {
+            pNextEntry = GetEOS();
+            return 1L;
+        }
+        
+        pNextEntry = NULL;
+        return E_BUFFER_NOT_FULL;
+    }
+        
+    pNextEntry = pNextCluster->GetFirst();
+    
+    while (pNextEntry)
+    {
+        const Block* const pNextBlock = pNextEntry->GetBlock();
+        assert(pNextBlock);
+        
+        if (pNextBlock->GetTrackNumber() == (unsigned long)m_info.number)
+            return 0L;
+            
+        pNextEntry = pNextCluster->GetNext(pNextEntry);
+    }
+    
+    //TODO: what has happened here is that we did not find a block
+    //with a matching track number on the next cluster.  It might
+    //be the case that some cluster beyond the next cluster 
+    //contains a block having a matching track number, but for
+    //now we terminate the search immediately.  We do this so that
+    //we don't end up searching the entire file looking for the
+    //next block.  Another possibility is to try searching for the next
+    //block in a small, fixed number of clusters (intead searching
+    //just the next one), or to terminate the search when when the
+    //there is a large gap in time, or large gap in file position.  It
+    //might very well be the case that the approach we use here is
+    //unnecessarily conservative.
+    
+    //TODO: again, here's a case where we need to return the special
+    //EOS block.  Or something.  It's OK if pNext is NULL, because
+    //we only need it to set the stop time of the media sample.
+    //(The start time is determined from pCurr, which is non-NULL
+    //and non-EOS.)  The problem is when we set pCurr=pNext; when
+    //pCurr has the value NULL we interpret that to mean that we
+    //haven't fully initialized pCurr and we attempt to set it to
+    //point to the first block for this track.  But that's not what
+    //we want at all; we want the next call to PopulateSample to
+    //return end-of-stream, not (re)start from the beginning.
+    //
+    //One work-around is to send EOS immediately.  We would send 
+    //the EOS the next pass anyway, so maybe it's no great loss.  The 
+    //only problem is that if this the stream really does end one
+    //cluster early (relative to other tracks), or the last frame
+    //happens to be a keyframe ("CanSeekToEnd").
+    //
+    //The problem is that we need a way to mark as stream as
+    //"at end of stream" without actually being at end of stream.
+    //We need to give pCurr some value that means "you've reached EOS".
+    //We can't synthesize the special EOS Cluster immediately
+    //(when we first open the file, say), because we use the existance
+    //of that special cluster value to mean that we've read all of 
+    //the clusters (this is a network download, so we can't know apriori
+    //how many we have).
+    //
+    //Or, we could return E_FAIL, and set another bit in the stream
+    //object itself, to indicate that it should send EOS earlier
+    //than when (pCurr=pStop).
+    //
+    //Or, probably the best solution, when we actually load the 
+    //blocks into a cluster: if we notice that there's no block
+    //for a track, we synthesize a nonce EOS block for that track.
+    //That way we always have something to return.  But that will
+    //only work for sequential scan???
+
+    //pNext = NULL;    
+    //return E_FAIL;
+    pNextEntry = GetEOS();
+    return 1L;
+}
+
+
+Track::EOSBlock::EOSBlock()
+{
+}
+
+
+bool Track::EOSBlock::EOS() const
+{
+    return true;
+}
+
+
+Cluster* Track::EOSBlock::GetCluster() const
+{
+    return NULL;
+}
+
+
+size_t Track::EOSBlock::GetIndex() const
+{
+    return 0;
+}
+
+
+const Block* Track::EOSBlock::GetBlock() const
+{
+    return NULL;
+}
+
+
+bool Track::EOSBlock::IsBFrame() const
+{
+    return false;
+}
+
+
+VideoTrack::VideoTrack(Segment* pSegment, const Info& i) :
+    Track(pSegment, i),
+    m_width(-1),
+    m_height(-1),
+    m_rate(-1)
+{
+    assert(i.type == 1);
+    assert(i.number > 0);
+    
+    IMkvReader* const pReader = pSegment->m_pReader;
+    
+    const Settings& s = i.settings;
+    assert(s.start >= 0);
+    assert(s.size >= 0);
+    
+    long long pos = s.start;
+    assert(pos >= 0);
+    
+    const long long stop = pos + s.size;
+    
+    while (pos < stop)
+    {
+#ifdef _DEBUG
+        long len;
+        const long long id = ReadUInt(pReader, pos, len);
+        assert(id >= 0);  //TODO: handle error case
+        assert((pos + len) <= stop);
+#endif
+        if (Match(pReader, pos, 0x30, m_width))         
+            ;
+        else if (Match(pReader, pos, 0x3A, m_height))   
+            ;
+        else if (Match(pReader, pos, 0x0383E3, m_rate)) 
+            ;
+        else
+        {
+            long len;
+            const long long id = ReadUInt(pReader, pos, len);
+            assert(id >= 0);  //TODO: handle error case
+            assert((pos + len) <= stop);
+        
+            pos += len;  //consume id
+            
+            const long long size = ReadUInt(pReader, pos, len);
+            assert(size >= 0);  //TODO: handle error case
+            assert((pos + len) <= stop);
+            
+            pos += len;  //consume length of size
+            assert((pos + size) <= stop);
+            
+            //pos now designates start of payload
+            
+            pos += size;  //consume payload
+            assert(pos <= stop);
+        }
+    }
+    
+    return;
+}
+
+
+bool VideoTrack::VetEntry(const BlockEntry* pBlockEntry) const
+{
+    assert(pBlockEntry);
+    
+    const Block* const pBlock = pBlockEntry->GetBlock();
+    assert(pBlock);    
+    assert(pBlock->GetTrackNumber() == (unsigned long)m_info.number);
+    
+    return pBlock->IsKey();
+}
+
+
+
+long long VideoTrack::GetWidth() const
+{
+    return m_width;
+}
+
+
+long long VideoTrack::GetHeight() const
+{
+    return m_height;
+}
+
+
+double VideoTrack::GetFrameRate() const
+{
+    return m_rate;
+}
+
+
+AudioTrack::AudioTrack(Segment* pSegment, const Info& i) :
+    Track(pSegment, i)
+{
+    assert(i.type == 2);
+    assert(i.number > 0);
+
+    IMkvReader* const pReader = pSegment->m_pReader;
+    
+    const Settings& s = i.settings;
+    assert(s.start >= 0);
+    assert(s.size >= 0);
+    
+    long long pos = s.start;
+    assert(pos >= 0);
+    
+    const long long stop = pos + s.size;
+    
+    while (pos < stop)
+    {
+#ifdef _DEBUG
+        long len;
+        const long long id = ReadUInt(pReader, pos, len);
+        assert(id >= 0);  //TODO: handle error case
+        assert((pos + len) <= stop);
+#endif
+        if (Match(pReader, pos, 0x35, m_rate))            
+            ;
+        else if (Match(pReader, pos, 0x1F, m_channels))   
+            ;
+        else if (Match(pReader, pos, 0x2264, m_bitDepth))  
+            ;            
+        else
+        {
+            long len;
+            const long long id = ReadUInt(pReader, pos, len);
+            assert(id >= 0);  //TODO: handle error case
+            assert((pos + len) <= stop);
+        
+            pos += len;  //consume id
+            
+            const long long size = ReadUInt(pReader, pos, len);
+            assert(size >= 0);  //TODO: handle error case
+            assert((pos + len) <= stop);
+            
+            pos += len;  //consume length of size
+            assert((pos + size) <= stop);
+            
+            //pos now designates start of payload
+            
+            pos += size;  //consume payload
+            assert(pos <= stop);
+        }
+    }
+
+    return;
+}
+
+bool AudioTrack::VetEntry(const BlockEntry* pBlockEntry) const
+{
+    assert(pBlockEntry);
+    
+    const Block* const pBlock = pBlockEntry->GetBlock();
+    assert(pBlock);
+    assert(pBlock->GetTrackNumber() == (unsigned long)m_info.number);
+
+    return true;
+}
+
+
+double AudioTrack::GetSamplingRate() const
+{
+    return m_rate;
+}
+
+
+long long AudioTrack::GetChannels() const
+{
+    return m_channels;
+}
+
+long long AudioTrack::GetBitDepth() const
+{
+    return m_bitDepth;
+}
+
+Tracks::Tracks(Segment* pSegment, long long start, long long size_) :
+    m_pSegment(pSegment),
+    m_start(start),
+    m_size(size_),
+    m_trackEntries(NULL),
+    m_trackEntriesEnd(NULL)
+{
+    long long stop = m_start + m_size;
+    IMkvReader* const pReader = m_pSegment->m_pReader;
+    
+    long long pos1 = m_start;
+    int count = 0;
+    
+    while (pos1 < stop)
+    {
+        long len;
+        const long long id = ReadUInt(pReader, pos1, len);
+        assert(id >= 0);
+        assert((pos1 + len) <= stop);
+        
+        pos1 += len;  //consume id
+        
+        const long long size = ReadUInt(pReader, pos1, len);
+        assert(size >= 0);
+        assert((pos1 + len) <= stop);
+        
+        pos1 += len;  //consume length of size
+        
+        //pos now desinates start of element
+        if (id == 0x2E)  //TrackEntry ID
+            ++count;
+            
+        pos1 += size;  //consume payload
+        assert(pos1 <= stop);
+    }    
+
+    if (count <= 0)
+        return;
+
+    m_trackEntries = new Track*[count];
+    m_trackEntriesEnd = m_trackEntries;
+
+    long long pos = m_start;
+
+    while (pos < stop)
+    {
+        long len;
+        const long long id = ReadUInt(pReader, pos, len);
+        assert(id >= 0);
+        assert((pos + len) <= stop);
+        
+        pos += len;  //consume id
+        
+        const long long size1 = ReadUInt(pReader, pos, len);
+        assert(size1 >= 0);
+        assert((pos + len) <= stop);
+        
+        pos += len;  //consume length of size
+        
+        //pos now desinates start of element
+        
+        if (id == 0x2E)  //TrackEntry ID
+            ParseTrackEntry(pos, size1, *m_trackEntriesEnd++);
+            
+        pos += size1;  //consume payload
+        assert(pos <= stop);
+    }    
+}
+
+unsigned long Tracks::GetTracksCount() const
+{
+    const ptrdiff_t result = m_trackEntriesEnd - m_trackEntries;
+    assert(result >= 0);
+    
+    return static_cast<unsigned long>(result);
+}
+
+
+void Tracks::ParseTrackEntry(
+    long long start,
+    long long size,
+    Track*& pTrack)
+{
+    IMkvReader* const pReader = m_pSegment->m_pReader;
+    
+    long long pos = start;
+    const long long stop = start + size;
+
+    Track::Info i;
+    
+    Track::Settings videoSettings;
+    videoSettings.start = -1;
+    
+    Track::Settings audioSettings;
+    audioSettings.start = -1;
+    
+    while (pos < stop)
+    {
+#ifdef _DEBUG
+        long len;
+        const long long id = ReadUInt(pReader, pos, len);
+        len;
+        id;
+#endif
+        if (Match(pReader, pos, 0x57, i.number))
+            assert(i.number > 0);
+
+        else if (Match(pReader, pos, 0x33C5, i.uid))           
+            ;  
+
+        else if (Match(pReader, pos, 0x03, i.type))            
+            ;  
+
+        else if (Match(pReader, pos, 0x136E, i.nameAsUTF8))          
+            assert(i.nameAsUTF8);  
+
+        else if (Match(pReader, pos, 0x06, i.codecId))         
+            ;  
+
+        else if (Match(pReader, pos, 0x23A2, i.codecPrivate, &i.codecPrivateSize))  
+            ;  
+
+        else if (Match(pReader, pos, 0x058688, i.codecNameAsUTF8))   
+            assert(i.codecNameAsUTF8);  
+
+        else
+        {
+            long len;
+            
+            const long long id = ReadUInt(pReader, pos, len);
+            assert(id >= 0);  //TODO: handle error case
+            assert((pos + len) <= stop);
+            
+            pos += len;  //consume id
+            
+            const long long size = ReadUInt(pReader, pos, len);
+            assert(size >= 0);  //TODO: handle error case
+            assert((pos + len) <= stop);
+            
+            pos += len;  //consume length of size
+            const long long start = pos;
+            
+            pos += size;  //consume payload
+            assert(pos <= stop);
+            
+            if (id == 0x60)
+            {
+                videoSettings.start = start;
+                videoSettings.size = size;
+            }
+            else if (id == 0x61)
+            {
+                audioSettings.start = start;
+                audioSettings.size = size;
+            }
+        }
+    }
+    
+    assert(pos == stop);
+    //TODO: propertly vet info.number, to ensure both its existence,
+    //and that it is unique among all tracks.
+    assert(i.number > 0);
+
+    //TODO: vet settings, to ensure that video settings (0x60)
+    //were specified when type = 1, and that audio settings (0x61)
+    //were specified when type = 2.    
+    if (i.type == 1)  //video
+    {
+        assert(audioSettings.start < 0);
+        assert(videoSettings.start >= 0);
+        
+        i.settings = videoSettings;
+        
+        VideoTrack* const t = new VideoTrack(m_pSegment, i);
+        assert(t);  //TODO
+        pTrack = t;    
+    }
+    else if (i.type == 2)  //audio
+    {
+        assert(videoSettings.start < 0);
+        assert(audioSettings.start >= 0);
+        
+        i.settings = audioSettings;
+        
+        AudioTrack* const t = new  AudioTrack(m_pSegment, i);
+        assert(t);  //TODO
+        pTrack = t;  
+    }
+    else
+    {
+        // for now we do not support other track types yet.
+        // TODO: support other track types
+        i.Clear();
+  
+        pTrack = NULL;
+    }
+    
+    return;
+}
+
+
+Tracks::~Tracks()
+{
+    Track** i = m_trackEntries;
+    Track** const j = m_trackEntriesEnd;
+    
+    while (i != j)
+    {
+        Track* pTrack = *i++;
+        delete pTrack;
+        pTrack = NULL;    
+    }
+
+    delete[] m_trackEntries;
+}
+
+
+Track* Tracks::GetTrackByNumber(unsigned long tn) const
+{
+    Track** i = m_trackEntries;
+    Track** const j = m_trackEntriesEnd;
+
+    while (i != j)
+    {
+        Track* const pTrack = *i++;
+       
+        if (pTrack == NULL)
+            continue;
+
+        if (tn == pTrack->GetNumber())
+            return pTrack;
+    }
+
+    return NULL;  //not found
+}
+
+
+Track* Tracks::GetTrackByIndex(unsigned long idx) const
+{
+    const ptrdiff_t count = m_trackEntriesEnd - m_trackEntries;
+       
+    if (idx >= static_cast<unsigned long>(count))
+         return NULL;
+
+    return m_trackEntries[idx];
+}
+
+
+void Cluster::Load()
+{
+    assert(m_pSegment);
+    
+    if (m_start > 0)
+    {
+        assert(m_size > 0);
+        assert(m_timecode >= 0);
+        return;
+    }
+    
+    assert(m_size == 0);
+    assert(m_timecode < 0);
+    
+    IMkvReader* const pReader = m_pSegment->m_pReader;
+
+    const long long off = -m_start;  //relative to segment
+    long long pos = m_pSegment->m_start + off;  //absolute
+    
+    long len;
+
+    const long long id_ = ReadUInt(pReader, pos, len);
+    assert(id_ >= 0);
+    assert(id_ == 0x0F43B675);  //Cluster ID
+    
+    pos += len;  //consume id
+    
+    const long long size_ = ReadUInt(pReader, pos, len);
+    assert(size_ >= 0);
+    
+    pos += len;  //consume size
+    
+    m_start = pos;
+    m_size = size_;
+    
+    const long long stop = m_start + size_;
+    
+    long long timecode = -1;
+    
+    while (pos < stop)
+    {
+        if (Match(pReader, pos, 0x67, timecode))
+            break;            
+        else
+        {
+            const long long id = ReadUInt(pReader, pos, len);
+            assert(id >= 0);  //TODO
+            assert((pos + len) <= stop);
+            
+            pos += len;  //consume id
+            
+            const long long size = ReadUInt(pReader, pos, len);
+            assert(size >= 0);  //TODO
+            assert((pos + len) <= stop);
+            
+            pos += len;  //consume size
+            
+            if (id == 0x20)  //BlockGroup ID
+                break;
+                
+            if (id == 0x23)  //SimpleBlock ID
+                break;
+
+            pos += size;  //consume payload
+            assert(pos <= stop);
+        }
+    }
+    
+    assert(pos <= stop);
+    assert(timecode >= 0);
+    
+    m_timecode = timecode;
+}
+
+
+Cluster* Cluster::Parse(
+    Segment* pSegment,
+    size_t idx,
+    long long off)
+{
+    assert(pSegment);
+    assert(off >= 0);
+    assert(off < pSegment->m_size);
+    Cluster* const pCluster = new Cluster(pSegment, idx, -off);
+    assert(pCluster);
+    
+    return pCluster;
+}
+
+
+Cluster::Cluster() :
+    m_pSegment(NULL),
+    m_index(0),
+    m_start(0),
+    m_size(0),
+    m_timecode(0),
+    m_pEntries(NULL),
+    m_entriesCount(0)
+{
+}
+
+Cluster::Cluster(
+    Segment* pSegment,
+    size_t idx,
+    long long off) :
+    m_pSegment(pSegment),
+    m_index(idx),
+    m_start(off),
+    m_size(0),
+    m_timecode(-1),
+    m_pEntries(NULL),
+    m_entriesCount(0)
+{
+}
+
+
+Cluster::~Cluster()
+{
+#if 0
+    while (!m_pEntries.empty())
+    {
+        BlockEntry* pBlockEntry = m_pEntries.front();
+        assert(pBlockEntry);
+        
+        m_pEntries.pop_front();
+        delete pBlockEntry;
+    }
+#else
+    BlockEntry** i = m_pEntries;
+    BlockEntry** const j = m_pEntries + m_entriesCount;
+    while (i != j)
+    {
+         BlockEntry* p = *i++;
+   
+         assert(p);
+         delete p;
+    }
+ 
+    delete[] m_pEntries;
+#endif
+
+}
+
+bool Cluster::EOS() const
+{
+    return (m_pSegment == 0);
+}
+
+
+void Cluster::LoadBlockEntries()
+{
+    if (m_pEntries)
+        return;
+
+    Load();    
+    assert(m_timecode >= 0);
+    assert(m_start > 0);
+    assert(m_size > 0);
+    
+    IMkvReader* const pReader = m_pSegment->m_pReader;
+    
+    long long pos = m_start;
+    const long long stop = m_start + m_size;
+    long long timecode = -1;
+   
+    long long idx = pos;
+
+    m_entriesCount = 0;
+    
+    while (idx < stop)
+    {
+        if (Match(pReader, idx, 0x67, timecode))
+            assert(timecode == m_timecode);
+        else 
+        {
+            long len;
+            
+            const long long id = ReadUInt(pReader, idx, len);
+            assert(id >= 0);  //TODO
+            assert((idx + len) <= stop);
+            
+            idx += len;  //consume id
+            
+            const long long size = ReadUInt(pReader, idx, len);
+            assert(size >= 0);  //TODO
+            assert((idx + len) <= stop);
+            
+            idx += len;  //consume size
+            
+            if (id == 0x20)  //BlockGroup ID
+                ++m_entriesCount;
+            else if (id == 0x23)  //SimpleBlock ID
+                ++m_entriesCount;
+
+            idx += size;  //consume payload
+
+            assert(idx <= stop);
+        }  
+    }
+
+    if (m_entriesCount == 0)
+        return;
+     
+    m_pEntries = new BlockEntry*[m_entriesCount];
+    size_t index = 0;
+    
+    while (pos < stop)
+    {
+        if (Match(pReader, pos, 0x67, timecode))
+            assert(timecode == m_timecode);
+        else
+        {
+            long len;
+            const long long id = ReadUInt(pReader, pos, len);
+            assert(id >= 0);  //TODO
+            assert((pos + len) <= stop);
+            
+            pos += len;  //consume id
+            
+            const long long size = ReadUInt(pReader, pos, len);
+            assert(size >= 0);  //TODO
+            assert((pos + len) <= stop);
+            
+            pos += len;  //consume size
+            
+            if (id == 0x20)  //BlockGroup ID
+                ParseBlockGroup(pos, size, index++);
+            else if (id == 0x23)  //SimpleBlock ID
+                ParseSimpleBlock(pos, size, index++);
+
+            pos += size;  //consume payload
+            assert(pos <= stop);
+        }
+    }
+    
+    assert(pos == stop);
+    assert(timecode >= 0);
+    assert(index == m_entriesCount);
+}
+
+
+
+long long Cluster::GetTimeCode()
+{
+    Load();
+    return m_timecode;
+}
+
+
+long long Cluster::GetTime()
+{
+    const long long tc = GetTimeCode();
+    assert(tc >= 0);
+    
+    const SegmentInfo* const pInfo = m_pSegment->GetInfo();
+    assert(pInfo);
+    
+    const long long scale = pInfo->GetTimeCodeScale();
+    assert(scale >= 1);
+    
+    const long long t = m_timecode * scale;
+
+    return t;
+}
+
+
+void Cluster::ParseBlockGroup(long long start, long long size, size_t index)
+{
+    assert(m_pEntries);
+    assert(m_entriesCount);
+    assert(index < m_entriesCount);
+    
+    BlockGroup* const pGroup = new BlockGroup(this, index, start, size);
+    assert(pGroup);  //TODO
+        
+    m_pEntries[index] = pGroup;
+}
+
+
+
+void Cluster::ParseSimpleBlock(long long start, long long size, size_t index)
+{
+    assert(m_pEntries);
+    assert(m_entriesCount);
+    assert(index < m_entriesCount);
+
+    SimpleBlock* const pSimpleBlock = new SimpleBlock(this, index, start, size);
+    assert(pSimpleBlock);  //TODO
+        
+    m_pEntries[index] = pSimpleBlock;
+}
+
+
+const BlockEntry* Cluster::GetFirst()
+{
+    LoadBlockEntries();
+    
+    return m_pEntries[0];
+}
+
+        
+const BlockEntry* Cluster::GetLast()
+{ 
+    if (m_entriesCount == 0)
+        return m_pEntries[0];
+    
+    return m_pEntries[m_entriesCount-1];
+}
+
+        
+const BlockEntry* Cluster::GetNext(const BlockEntry* pEntry) const
+{
+    assert(pEntry);
+    
+    size_t idx = pEntry->GetIndex();
+    
+    ++idx;
+
+    if (idx == m_entriesCount) 
+      return NULL;
+
+    return m_pEntries[idx];
+
+}
+
+
+const BlockEntry* Cluster::GetEntry(const Track* pTrack)
+{
+
+    assert(pTrack);
+    
+    if (m_pSegment == NULL)  //EOS
+        return pTrack->GetEOS();
+    
+    LoadBlockEntries();
+    
+    BlockEntry* i = *m_pEntries;
+    BlockEntry* j = *m_pEntries + m_entriesCount;
+    while (i != j)
+    {
+        BlockEntry* pEntry = i;
+        i++;
+        assert(pEntry);
+        assert(!pEntry->EOS());
+        
+        const Block* const pBlock = pEntry->GetBlock();
+        assert(pBlock);
+        
+        if (pBlock->GetTrackNumber() != pTrack->GetNumber())
+            continue;
+
+        if (pTrack->VetEntry(pEntry))
+            return pEntry;
+    }
+    
+    return pTrack->GetEOS();  //no satisfactory block found
+}
+
+
+BlockEntry::BlockEntry()
+{
+}
+
+
+BlockEntry::~BlockEntry()
+{
+}
+
+
+
+SimpleBlock::SimpleBlock(
+    Cluster* pCluster, 
+    size_t idx, 
+    long long start, 
+    long long size) :
+    m_pCluster(pCluster),
+    m_index(idx),
+    m_block(start, size, pCluster->m_pSegment->m_pReader)
+{
+}
+
+
+bool SimpleBlock::EOS() const
+{
+    return false;
+}
+
+
+Cluster* SimpleBlock::GetCluster() const
+{
+    return m_pCluster;
+}
+
+
+size_t SimpleBlock::GetIndex() const
+{
+    return m_index;
+}
+
+
+const Block* SimpleBlock::GetBlock() const
+{
+    return &m_block;
+}
+
+
+bool SimpleBlock::IsBFrame() const
+{
+    return false;
+}
+
+
+BlockGroup::BlockGroup(
+    Cluster* pCluster, 
+    size_t idx, 
+    long long start, 
+    long long size_) :
+    m_pCluster(pCluster),
+    m_index(idx),
+    m_prevTimeCode(0),
+    m_nextTimeCode(0),
+    m_pBlock(NULL)  //TODO: accept multiple blocks within a block group
+{
+    IMkvReader* const pReader = m_pCluster->m_pSegment->m_pReader;
+    
+    long long pos = start;
+    const long long stop = start + size_;
+ 
+    bool bSimpleBlock = false;
+    
+    while (pos < stop)
+    {
+        short t;
+    
+        if (Match(pReader, pos, 0x7B, t))
+        {    
+            if (t < 0)
+                m_prevTimeCode = t;
+            else if (t > 0)
+                m_nextTimeCode = t;
+            else
+                assert(false);
+        }
+        else
+        {
+            long len;
+            const long long id = ReadUInt(pReader, pos, len);
+            assert(id >= 0);  //TODO
+            assert((pos + len) <= stop);
+            
+            pos += len;  //consume ID
+            
+            const long long size = ReadUInt(pReader, pos, len);
+            assert(size >= 0);  //TODO
+            assert((pos + len) <= stop);
+            
+            pos += len;  //consume size
+            
+            switch (id)
+            {
+                case 0x23:  //SimpleBlock ID
+                    bSimpleBlock = true;
+                    //YES, FALL THROUGH TO NEXT CASE
+
+                case 0x21:  //Block ID
+                    ParseBlock(pos, size);                    
+                    break;
+                    
+                default:
+                    break;
+            }
+                
+            pos += size;  //consume payload
+            assert(pos <= stop);
+        }
+    }
+    
+    assert(pos == stop);
+    assert(m_pBlock);
+    
+    if (!bSimpleBlock)
+        m_pBlock->SetKey(m_prevTimeCode >= 0);
+}
+
+
+BlockGroup::~BlockGroup()
+{
+    delete m_pBlock;
+}
+
+
+void BlockGroup::ParseBlock(long long start, long long size)
+{   
+    IMkvReader* const pReader = m_pCluster->m_pSegment->m_pReader;
+    
+    Block* const pBlock = new Block(start, size, pReader);
+    assert(pBlock);  //TODO
+
+    //TODO: the Matroska spec says you have multiple blocks within the 
+    //same block group, with blocks ranked by priority (the flag bits).
+    //I haven't ever seen such a file (mkvmux certainly doesn't make
+    //one), so until then I'll just assume block groups contain a single
+    //block.
+#if 0    
+    m_blocks.push_back(pBlock);
+#else
+    assert(m_pBlock == NULL);
+    m_pBlock = pBlock;
+#endif
+
+#if 0
+    Track* const pTrack = pBlock->GetTrack();
+    assert(pTrack);
+    
+    pTrack->Insert(pBlock);
+#endif
+}
+
+
+bool BlockGroup::EOS() const
+{
+    return false;
+}
+
+
+Cluster* BlockGroup::GetCluster() const
+{
+    return m_pCluster;
+}
+
+
+size_t BlockGroup::GetIndex() const
+{
+    return m_index;
+}
+
+
+const Block* BlockGroup::GetBlock() const
+{
+    return m_pBlock;
+}
+
+
+short BlockGroup::GetPrevTimeCode() const
+{
+    return m_prevTimeCode;
+}
+
+
+short BlockGroup::GetNextTimeCode() const
+{
+    return m_nextTimeCode;
+}    
+
+
+bool BlockGroup::IsBFrame() const
+{
+    return (m_nextTimeCode > 0);
+}
+
+
+
+Block::Block(long long start, long long size_, IMkvReader* pReader) :
+    m_start(start),
+    m_size(size_)
+{
+    long long pos = start;
+    const long long stop = start + size_;
+
+    long len;
+    
+    m_track = ReadUInt(pReader, pos, len);
+    assert(m_track > 0);
+    assert((pos + len) <= stop);
+    
+    pos += len;  //consume track number
+    assert((stop - pos) >= 2);
+    
+    m_timecode = Unserialize2SInt(pReader, pos);
+
+    pos += 2;
+    assert((stop - pos) >= 1);
+    
+    const long hr = pReader->Read(pos, 1, &m_flags);
+    assert(hr == 0L);
+
+    ++pos;
+    assert(pos <= stop);
+    
+    m_frameOff = pos;
+    
+    const long long frame_size = stop - pos;
+
+    assert(frame_size <= 2147483647L);
+    
+    m_frameSize = static_cast<long>(frame_size);
+}
+
+
+long long Block::GetTimeCode(Cluster* pCluster) const
+{
+    assert(pCluster);
+    
+    const long long tc0 = pCluster->GetTimeCode();
+    assert(tc0 >= 0);
+    
+    const long long tc = tc0 + static_cast<long long>(m_timecode);
+    assert(tc >= 0);
+    
+    return tc;  //unscaled timecode units
+}
+
+
+long long Block::GetTime(Cluster* pCluster) const
+{
+    assert(pCluster);
+    
+    const long long tc = GetTimeCode(pCluster);
+    
+    const Segment* const pSegment = pCluster->m_pSegment;
+    const SegmentInfo* const pInfo = pSegment->GetInfo();
+    assert(pInfo);
+    
+    const long long scale = pInfo->GetTimeCodeScale();
+    assert(scale >= 1);
+    
+    const long long ns = tc * scale;
+
+    return ns;
+}
+
+
+unsigned long Block::GetTrackNumber() const
+{
+    assert(m_track > 0);
+    
+    return static_cast<unsigned long>(m_track);
+}
+
+
+bool Block::IsKey() const
+{
+    return ((m_flags & static_cast<unsigned char>(1 << 7)) != 0);
+}
+
+
+void Block::SetKey(bool bKey)
+{
+    if (bKey)
+        m_flags |= static_cast<unsigned char>(1 << 7);
+    else
+        m_flags &= 0x7F;
+}
+
+
+long Block::GetSize() const
+{
+    return m_frameSize;
+}
+
+
+long Block::Read(IMkvReader* pReader, unsigned char* buf) const
+{
+
+    assert(pReader);
+    assert(buf);
+    
+    const long hr = pReader->Read(m_frameOff, m_frameSize, buf);
+    
+    return hr;
+}
+
+
+}  //end namespace mkvparser
diff --git a/media/libstagefright/matroska/mkvparser.hpp b/media/libstagefright/matroska/mkvparser.hpp
new file mode 100644
index 0000000..4d311b4
--- /dev/null
+++ b/media/libstagefright/matroska/mkvparser.hpp
@@ -0,0 +1,428 @@
+#ifndef MKVPARSER_HPP
+#define MKVPARSER_HPP
+
+#include <cstdlib>
+#include <cstdio>
+
+namespace mkvparser
+{
+
+const int E_FILE_FORMAT_INVALID = -2;
+const int E_BUFFER_NOT_FULL = -3;
+
+class IMkvReader
+{
+public:
+    virtual int Read(long long position, long length, unsigned char* buffer) = 0;
+    virtual int Length(long long* total, long long* available) = 0; 	
+protected:
+    virtual ~IMkvReader();
+};
+
+long long GetUIntLength(IMkvReader*, long long, long&);
+long long ReadUInt(IMkvReader*, long long, long&);
+long long SyncReadUInt(IMkvReader*, long long pos, long long stop, long&);
+long long UnserializeUInt(IMkvReader*, long long pos, long long size);
+float Unserialize4Float(IMkvReader*, long long);
+double Unserialize8Double(IMkvReader*, long long);
+short Unserialize2SInt(IMkvReader*, long long);
+signed char Unserialize1SInt(IMkvReader*, long long);
+bool Match(IMkvReader*, long long&, unsigned long, long long&);
+bool Match(IMkvReader*, long long&, unsigned long, char*&);
+bool Match(IMkvReader*, long long&, unsigned long,unsigned char*&,
+           size_t *optionalSize = NULL);
+bool Match(IMkvReader*, long long&, unsigned long, double&);
+bool Match(IMkvReader*, long long&, unsigned long, short&);
+
+
+struct EBMLHeader
+{
+    EBMLHeader();
+    ~EBMLHeader();  
+    long long m_version;
+    long long m_readVersion;
+    long long m_maxIdLength;
+    long long m_maxSizeLength;
+    char* m_docType;
+    long long m_docTypeVersion;
+    long long m_docTypeReadVersion;
+    
+    long long Parse(IMkvReader*, long long&);
+};
+
+
+class Segment;
+class Track;
+class Cluster;
+
+class Block
+{
+    Block(const Block&);
+    Block& operator=(const Block&);
+
+public:
+    const long long m_start;
+    const long long m_size;
+    
+    Block(long long start, long long size, IMkvReader*);
+    
+    unsigned long GetTrackNumber() const;
+    
+    long long GetTimeCode(Cluster*) const;  //absolute, but not scaled
+    long long GetTime(Cluster*) const;      //absolute, and scaled (nanosecond units)
+    bool IsKey() const;
+    void SetKey(bool);
+
+    long GetSize() const;
+    long Read(IMkvReader*, unsigned char*) const;
+    
+private:
+    long long m_track;   //Track::Number()
+    short m_timecode;  //relative to cluster
+    unsigned char m_flags;
+    long long m_frameOff;
+    long m_frameSize;    
+
+};
+
+
+class BlockEntry
+{
+    BlockEntry(const BlockEntry&);
+    BlockEntry& operator=(const BlockEntry&);
+    
+public:
+    virtual ~BlockEntry();
+    virtual bool EOS() const = 0;
+    virtual Cluster* GetCluster() const = 0;
+    virtual size_t GetIndex() const = 0;
+    virtual const Block* GetBlock() const = 0;
+    virtual bool IsBFrame() const = 0;
+    
+protected:
+    BlockEntry();
+
+};
+
+
+class SimpleBlock : public BlockEntry
+{
+    SimpleBlock(const SimpleBlock&);
+    SimpleBlock& operator=(const SimpleBlock&);
+
+public:
+    SimpleBlock(Cluster*, size_t, long long start, long long size);
+
+    bool EOS() const;
+    Cluster* GetCluster() const; 
+    size_t GetIndex() const;
+    const Block* GetBlock() const;
+    bool IsBFrame() const;
+
+protected:
+    Cluster* const m_pCluster;
+    const size_t m_index;
+    Block m_block;
+    
+};
+
+
+class BlockGroup : public BlockEntry
+{
+    BlockGroup(const BlockGroup&);
+    BlockGroup& operator=(const BlockGroup&);
+
+public:
+    BlockGroup(Cluster*, size_t, long long, long long);
+    ~BlockGroup();
+    
+    bool EOS() const;
+    Cluster* GetCluster() const; 
+    size_t GetIndex() const;
+    const Block* GetBlock() const;
+    bool IsBFrame() const;
+    
+    short GetPrevTimeCode() const;  //relative to block's time
+    short GetNextTimeCode() const;  //as above
+    
+protected:
+    Cluster* const m_pCluster;
+    const size_t m_index;
+    
+private:
+    BlockGroup(Cluster*, size_t, unsigned long);
+    void ParseBlock(long long start, long long size);
+
+    short m_prevTimeCode;
+    short m_nextTimeCode;
+    
+    //TODO: the Matroska spec says you can have multiple blocks within the 
+    //same block group, with blocks ranked by priority (the flag bits).
+    //For now we just cache a single block.
+#if 0
+    typedef std::deque<Block*> blocks_t;
+    blocks_t m_blocks;  //In practice should contain only a single element.
+#else
+    Block* m_pBlock;
+#endif
+    
+};
+
+
+class Track
+{
+    Track(const Track&);
+    Track& operator=(const Track&);
+
+public:    
+    Segment* const m_pSegment;
+    virtual ~Track();
+    
+    long long GetType() const; 
+    unsigned long GetNumber() const;
+    const char* GetNameAsUTF8() const;
+    const char* GetCodecNameAsUTF8() const;
+    const char* GetCodecId() const;
+    const unsigned char* GetCodecPrivate(
+            size_t *optionalSize = NULL) const;
+    
+    const BlockEntry* GetEOS() const;
+    
+    struct Settings 
+    {
+        long long start;
+        long long size;
+    };
+    
+    struct Info
+    {
+        long long type;
+        long long number;
+        long long uid;
+        char* nameAsUTF8;
+        char* codecId;
+        unsigned char* codecPrivate;
+        size_t codecPrivateSize;
+        char* codecNameAsUTF8;
+        Settings settings;
+        Info();
+        void Clear();
+    };
+    
+    long GetFirst(const BlockEntry*&) const;
+    long GetNext(const BlockEntry* pCurr, const BlockEntry*& pNext) const;
+    virtual bool VetEntry(const BlockEntry*) const = 0;
+        
+protected:
+    Track(Segment*, const Info&);        
+    const Info m_info;
+
+    class EOSBlock : public BlockEntry
+    {
+    public:
+        EOSBlock();
+
+        bool EOS() const;    
+        Cluster* GetCluster() const;
+        size_t GetIndex() const;
+        const Block* GetBlock() const;
+        bool IsBFrame() const;
+    };
+    
+    EOSBlock m_eos;
+    
+};
+
+
+class VideoTrack : public Track
+{
+    VideoTrack(const VideoTrack&);
+    VideoTrack& operator=(const VideoTrack&);
+    
+public:
+    VideoTrack(Segment*, const Info&);    
+    long long GetWidth() const;
+    long long GetHeight() const;
+    double GetFrameRate() const;
+    
+    bool VetEntry(const BlockEntry*) const;
+
+private:
+    long long m_width;
+    long long m_height;
+    double m_rate;
+    
+};
+
+
+class AudioTrack : public Track
+{
+    AudioTrack(const AudioTrack&);
+    AudioTrack& operator=(const AudioTrack&);
+
+public:
+    AudioTrack(Segment*, const Info&);    
+    double GetSamplingRate() const;
+    long long GetChannels() const;
+    long long GetBitDepth() const;    
+    bool VetEntry(const BlockEntry*) const;
+
+private:
+    double m_rate;
+    long long m_channels;
+    long long m_bitDepth;
+};
+
+
+class Tracks
+{
+    Tracks(const Tracks&);
+    Tracks& operator=(const Tracks&);
+
+public:
+    Segment* const m_pSegment;
+    const long long m_start;
+    const long long m_size;
+    
+    Tracks(Segment*, long long start, long long size);
+    virtual ~Tracks();
+
+    Track* GetTrackByNumber(unsigned long tn) const;
+    Track* GetTrackByIndex(unsigned long idx) const;
+    
+private:
+    Track** m_trackEntries; 
+    Track** m_trackEntriesEnd;
+
+    void ParseTrackEntry(long long, long long, Track*&);
+    
+public:
+    unsigned long GetTracksCount() const;
+};
+
+
+class SegmentInfo
+{
+    SegmentInfo(const SegmentInfo&);
+    SegmentInfo& operator=(const SegmentInfo&);
+    
+public:
+    Segment* const m_pSegment;
+    const long long m_start;
+    const long long m_size;
+    
+    SegmentInfo(Segment*, long long start, long long size);
+    ~SegmentInfo();
+    long long GetTimeCodeScale() const;
+    long long GetDuration() const;  //scaled
+    const char* GetMuxingAppAsUTF8() const;
+    const char* GetWritingAppAsUTF8() const;
+    const char* GetTitleAsUTF8() const;
+    
+private:
+    long long m_timecodeScale;
+    double m_duration;
+    char* m_pMuxingAppAsUTF8;
+    char* m_pWritingAppAsUTF8;
+    char* m_pTitleAsUTF8;
+};
+
+
+class Cluster
+{
+    Cluster(const Cluster&);
+    Cluster& operator=(const Cluster&);
+
+public:
+    Segment* const m_pSegment;
+    const size_t m_index;
+    
+public:    
+    static Cluster* Parse(Segment*, size_t, long long off);
+
+    Cluster();  //EndOfStream
+    ~Cluster();
+    
+    bool EOS() const;
+    
+    long long GetTimeCode();  //absolute, but not scaled
+    long long GetTime();      //absolute, and scaled (nanosecond units)
+
+    const BlockEntry* GetFirst();
+    const BlockEntry* GetLast();
+    const BlockEntry* GetNext(const BlockEntry*) const;    
+    const BlockEntry* GetEntry(const Track*);
+protected:    
+    Cluster(Segment*, size_t, long long off);
+    
+private:
+    long long m_start;
+    long long m_size;    
+    long long m_timecode;
+    BlockEntry** m_pEntries;
+    size_t m_entriesCount;
+   
+    void Load();
+    void LoadBlockEntries();
+    void ParseBlockGroup(long long, long long, size_t);
+    void ParseSimpleBlock(long long, long long, size_t);
+    
+};
+
+
+class Segment
+{
+    Segment(const Segment&);
+    Segment& operator=(const Segment&);
+
+private:
+    Segment(IMkvReader*, long long pos, long long size);
+
+public:
+    IMkvReader* const m_pReader;
+    const long long m_start;  //posn of segment payload
+    const long long m_size;   //size of segment payload
+    Cluster m_eos;  //TODO: make private?
+    
+    static long long CreateInstance(IMkvReader*, long long, Segment*&);
+    ~Segment();
+
+    //for big-bang loading (source filter)
+    long Load();
+
+    //for incremental loading (splitter)    
+    long long Unparsed() const;
+    long long ParseHeaders();
+    long ParseCluster(Cluster*&, long long& newpos) const;
+    bool AddCluster(Cluster*, long long);
+
+    Tracks* GetTracks() const;    
+    const SegmentInfo* const GetInfo() const;
+    long long GetDuration() const;
+    
+    //NOTE: this turned out to be too inefficient.
+    //long long Load(long long time_nanoseconds);
+
+    Cluster* GetFirst();
+    Cluster* GetLast();
+    unsigned long GetCount() const;
+    
+    Cluster* GetNext(const Cluster*);
+    Cluster* GetCluster(long long time_nanoseconds);
+    
+private:
+    long long m_pos;  //absolute file posn; what has been consumed so far    
+    SegmentInfo* m_pInfo;
+    Tracks* m_pTracks;    
+    Cluster** m_clusters;
+    size_t m_clusterCount;
+
+    void ParseSeekHead(long long pos, long long size, size_t*);
+    void ParseSeekEntry(long long pos, long long size, size_t*);
+    void ParseSecondarySeekHead(long long off, size_t*);
+};
+
+
+}  //end namespace mkvparser
+
+#endif  //MKVPARSER_HPP